Principles of Materials Characterization and Metrology 0198830262, 9780198830269

Table of contents :
Cover
Principles of Materials Characterization and Metrology
Copyright
Dedication
Preface
Contents
1: Introduction to Materials Characterization, Analysis, and Metrology
1.1 Microstructure, Characterization, and the Materials Engineering Tetrahedron
1.2 Examples of Characterization and Analysis
1.2.1 Ni-Based Superalloys: Ultrahigh Temperature Materials for Jet Engines
1.2.2 Unraveling the Structure of Deoxyribonucleic Acid (DNA)
1.2.3 Characterizing a Picasso Painting Reveals Hidden Secrets
1.2.4 Failure Analysis: Metallurgy of the RMS Titanic
1.2.5 Beneath Our Feet:Microstructure of Rocks and Minerals
1.2.6 Ceramic Materials: Sintering and Grain Boundary Phases
1.2.7 Microstructure and the Properties of Materials: An Engineering Example
1.3 Probes for Characterization and Analysis: An Overview
1.3.1 Probes and Signals
1.3.2 Probes Based on the Electromagnetic Spectrum and their Attributes
1.3.3 Wave–Particle Duality
1.3.4 Nature and Propagation of Electromagnetic Waves
1.3.5 Interactions of Probes with Matter and Criteria for Technique Selection
1.3.5.1 Penetration Depth and Mean Free Path Length
1.3.5.2 Resolution
1.3.5.3 Damage
1.3.5.4 Specimen Preparation or Requirements
1.4 Methods of Characterization: Spectroscopy, Diffraction, and Imaging
1.4.1 Spectroscopy: Absorption, Emission, and Transition Processes
1.4.1.1 Characteristic X-Ray Emission
1.4.1.2 Non-Radiative Auger Emission
1.4.2 Scattering and Diffraction
1.4.3 Imaging and Microscopy
1.4.4 Digital Imaging
1.4.4.1 Image Acquisition, Digitization, and Storage
1.4.4.2 Pre-Processing: Look-Up Tables, Histogram Equalization, Point, and Kernel Operations
1.4.4.3 Image Segmentation
1.4.5 In Situ Methods Across Spatial and Temporal Scales
1.5 Features of Materials Used for Characterization
Summary
Further Reading
References
Untitled
2: Atomic Structure and Spectra
2.1 Introduction
2.2 Atomic Structure
2.2.1 Bohr–Rutherford–Sommerfeld Model
2.2.2 Quantum Mechanical Model
2.3 Atomic Spectra: Transitions, Emissions, and Secondary Processes
2.3.1 Dipole Selection Rules and Allowed Transitions of Electrons in Atoms
2.3.2 Characteristic X-Ray Emissions and their Nomenclature
2.3.3 Non-Radiative Auger Electron Emission
2.3.4 Electron Photoemission
2.4 X-Rays as Probes: Generation and Transmission of X-Rays
2.4.1 Laboratory Sources and Methods of X-Ray Generation
2.4.2 X-Ray Absorption and Filtering
2.4.3 Synchrotron Sources of X-Ray Radiation
2.5 X-Rays as Signals: Core-Level Spectroscopy with X-Rays
2.5.1 Instrumentation for Detecting X-Rays
2.5.1.1 Wave Length Dispersion Spectrometer (WDS)
2.5.1.2 Energy Dispersive X-Ray Spectrometer (EDXS)
2.5.2 Chemical X-Ray Microanalysis
2.5.2.1 Photon Incidence: X-Ray Fluorescence Spectroscopy (XRF)
2.5.2.2 Electron Incidence:Electron Probe Microanalysis (EPMA)
2.5.2.3 Quantitative X-Ray Microanalysis
2.6 Surface Analysis: Spectroscopy with Electrons
2.6.1 Instrumentation for Surface Analysis with Electron Spectroscopy
2.6.1.1 Vacuum Chamber and Other Components
2.6.1.2 Electrons as Signals:Electron Energy Spectrometers and Analyzers
2.6.2 Auger Electron Spectroscopy
2.6.3 X-Ray Photoelectron Spectroscopy
2.6.4 Surface Compositional Analysis with AES and XPS
2.6.5 Comparison of AES and XPS
2.7 Select Applications
2.7.1 XRF Analysis of Dental and Medical Specimens
2.7.2 Environmental Science: Contamination in Ground Water Colloids
Summary
Further Reading
References
Untitled
3: Bonding and Spectra of Molecules and Solids
3.1 Introduction
3.2 Bonds and Bands
3.3 Interatomic Bonding in Solids
3.3.1 Ionic Bonding
3.3.2 Covalent Bonding
3.3.3 Metallic Bonding
3.4 Molecular Spectra
3.4.1 Vibrational and Rotational Modes
3.4.2 Ultraviolet and Visible Spectroscopy (UV-Vis)
3.4.3 Classical Model of Rayleigh and Raman Scattering
3.4.4 Selection Criteria for Infrared and Raman Activity
3.5 Infrared Spectroscopy
3.5.1 Instrumentation for Raman and IR Spectroscopy
3.5.2 Michelson Interferometer and the Fourier Transform Infrared (FTIR) Method
3.5.3 Practice and Application of FTIR
3.6 Raman Spectroscopy
3.6.1 Raman, Resonant Raman, and Fluorescence
3.6.2 Instrumentation for Raman Spectroscopy and Imaging
3.6.3 Application of Raman Spectroscopy in Chemical and Materials Analysis
3.6.4 Surface-Enhanced Raman Spectroscopy (SERS)
3.7 Probing the Electronic Structure of Solids
3.8 Photoemission and Inverse Photoemission from Solids
3.9 Absorption Spectroscopies–Probing Unoccupied States
3.9.1 X-Ray Absorption Spectroscopy (XAS)
3.9.2 Near-Edge and Extended X-Ray Absorption Fine Structure (NEXAFS and EXAFS)
3.10 Select Applications
3.10.1 Structure of Proteins Resolved by FTIR
3.10.2 Analysis of Catalytic Particles by XAS and XPS
Summary
Further Reading
References
Untitled
4: Crystallography and Diffraction
4.1 The Crystalline State
4.1.1 Lattices
4.1.2 Generalized Crystal Systems and Bravais Lattices
4.1.3 Lattice Points, Lines, Directions, and Planes
4.1.4 Zonal Equations
4.1.5 Atomic Size, Coordination, and Close Packing
4.1.6 Describing Crystal Structures—Some Examples
4.1.7 Symmetry and the International Tables for Crystallography
4.1.8 The Stereographic Projection
4.1.9 Imperfections in Crystals
4.1.9.1 Point Defects
4.1.9.2 Line Defects
4.1.9.3 Planar Defects
4.2 The Reciprocal Lattice
4.3 Diffraction
4.3.1 Bragg’s Law: Interpreting Diffraction in Real Space
4.3.2 The Ewald Construction: Interpreting Diffraction in Reciprocal Space
4.3.3 Comparison of X-Ray and Electron Diffraction
4.4 Quasicrystals and the Definition of a Crystalline Material
Summary
Further Reading
References
Untitled
5: Probes: Sources and Their Interactions with Matter
5.1 Introduction
5.2 Probes and Their Generation
5.2.1 Photons: Lamps and Lasers
5.2.1.1 Light Sources for Optical Microscopy
5.2.1.2 Simulated Emissions and Lasers
5.2.2 Electrons: Thermionic and Field-Emission Sources
5.2.3 Neutrons
5.2.4 Ions
5.2.4.1 Guiding and Focusing
5.3 Interactions of Probes with Matter, Including Damage
5.3.1 Photons
5.3.2 Electrons
5.3.2.1 Beam Broadening
5.3.2.2 Atomic Displacements
5.3.2.3 Sputtering
5.3.2.4 Beam Heating
5.3.2.5 Electrostatic Charging
5.3.2.6 Hydrocarbon Contamination
5.3.3 Neutrons
5.3.4 Protons
5.3.5 Ions
5.3.5.1 Dimensions, Sizes, and Cross-Sections
5.3.5.2 Ion–Solid Interactions
5.3.5.3 Physics of Kinematic Collisions
5.3.5.4 Specimen Damage
5.3.5.5 Ion-Beam Sputtering
5.4 Ion-Based Characterization Methods
5.4.1 Rutherford Back-Scattering Spectroscopy (RBS)
5.4.1.1 Energy Width in Back-Scattering Spectroscopy
5.4.1.2 Shape of the Back-Scattering Spectrum
5.4.1.3 Ni Thin Films Grown on Silicon: A Technological Example
5.4.2 Low-Energy Ion Scattering Spectroscopy (LEISS)
5.4.3 Secondary Ion Mass Spectrometry (SIMS)
5.4.4 Induction-Coupled Plasma Mass Spectrometry (ICP-MS)
5.4.4.1 Application of ICP-MS in the Pharmaceutical Industry
5.4.5 Particle-Induced X-Ray Emission (PIXE)
Summary
Further Reading
References
Untitled
6: Optics, Optical Methods, and Microscopy
6.1 Introduction
6.2 Wave Equation for Simple Harmonic Motion
6.2.1 The Phase Angle
6.2.2 The Superposition Principle
6.2.3 Phasor Representation and the Addition of Waves
6.2.4 Complex Representation of a Simple Harmonic Wave
6.2.5 Superposition of Two Waves of the Same Frequency
6.2.6 Addition of Waves on Orthogonal Planes and Polarization
6.3 Huygens’ Principle
6.4 Young Double-Slit Experiment
6.5 Reflection and Refraction
6.6 Diffraction
6.6.1 Fraunhofer Diffraction from a Single Slit
6.6.2 Fraunhofer Diffraction from Double and Multiple Slits
6.6.3 Resolving Power of a Diffraction Grating
6.6.4 Fresnel Diffraction
6.6.5 Fresnel Half-Period Zones
6.6.6 Diffraction by a Circular Aperture or Disc
6.6.7 Zone Plates and Their Applications in X-Ray Microscopy
6.7 Visually Observable: Characteristics of the Human Eye
6.8 Optical Microscopy
6.8.1 Resolution: Rayleigh and Abbe Criteria
6.8.2 Geometric Optics and Aberrations
6.8.2.1 Lens Defects: Aberrations, Distortions, and Astigmatism
6.8.2.2 Depth of Field and Depth of Focus
6.8.3 The Optical Microscope
6.8.3.1 Vertical Illumination in Reflection Geometry
6.8.3.2 Direct, Oblique, and Dark Field Imaging
6.8.3.3 Interference Contrast Microscopy
6.8.3.4 Optical Microscopy with Polarized Light
6.8.4 Confocal Scanning Optical Microscopy (CSOM)
6.8.5 Metallography
6.9 Ellipsometry
Unknown
exp (-µt).
6.9.1 pand s-Polarized Light Waves, and Fresnel Equations of Reflection
E, and magnetic
B, of pand s-polarized light upon reflection from a surface. The
E and B components of the light wave have to satisfy the boundary condition that
cos .i Erp cos .r = Etp cos .t (6.9.6a)
Eip Bip Brp Erp
Etp Btp
Eis Bis Brs Ers
Ets Bts
E, and magnetic induction,B, for (a) p-polarization and (b) s-polarization, at the interface
cos .i ni cos .t
cos .i + ni cos .t
cos .i + Brs cos .r = -Bts cos .t (6.9.8b)
cos .i nt cos .t
cos .i + nt cos .t
cos .i and Ntt = et eisin2.i 1/2
exp idrp (6.9.12)
exp(idrs) (6.9.13)
tan .B = nt/ni (6.9.14)
tan-1(1.49) = 56.. Further, .B
Example 6.9.1: Fused glass has a refractive index, n = 1.45. For a film of this material placed in air and with Brewster angle i
(6.9.14) Then, rp = Erp
And rs = Ers
6.9.2 Optical Elements Used in Ellipsometry
6.9.2.1 Polarizer (Analyzer)
6.9.2.2 Compensator (Retarder) and Photoelastic Modulator
Ex and Ey, due to
Ey Ey Ex Ex
6.9.3 Ellipsometry Measurements
Eip and Eis, optimally incident at the Brewster angle and
E Eis Eip Ers Erp
Summary
Further Reading
References
Untitled
7: X-Ray Diffraction
7.1 Introduction
7.2 Interaction of X-Rays with Electrons
7.2.1 Thomson Coherent Scattering
7.2.2 Compton Incoherent Scattering
7.3 Scattering by an Atom: Atomic Scattering Factor
7.4 Scattering by a Crystal: Structure Factor
7.5 Examples of Structure Factor Calculations
7.5.1 Face-Centered Cubic (FCC) Structure
7.5.2 Body-Centered Cubic (BCC) Structure
7.5.3 Hexagonal Close Packed (HCP) Structure
7.5.4 Cesium Chloride (CsCl) Structure
7.6 Symmetry and Structure Factor
7.6.1 Crystals with Inversion Symmetry
7.6.2 Friedel Law
7.6.3 Systematic Absences
7.7 The Inverse Problem of Determining Structure from Diffraction Intensities
7.8 Broadening of Diffracted Beams and Reciprocal Lattice Points
7.9 Methods of X-Ray Diffraction
7.9.1 The Laue Method for Single Crystals
7.9.2 Diffractometry of Powders and Single Crystals
7.9.3 Debye–Scherrer Method for Powders
7.9.4 Thin Films and Multilayers: Diffractometry, Reflectivity, and Pole Figures
7.9.5 Practical Considerations: Collimators and Monochromators
7.9.5.1 Collimators
7.9.5.2 Monochromators
7.10 Factors Influencing X-Ray Diffraction Intensities
7.10.1 Temperature Factor
7.10.2 Absorption or Transmission Factor
7.10.3 Lorentz Polarization Factor
7.10.4 Multiplicity
7.10.5 Corrected Intensities for Diffractometry and the Debye–Scherrer Camera
7.11 Applications of X-Ray Diffraction
7.11.1 Measurement of Lattice Parameters
7.11.2 Crystallite or Grain Size and Lattice Strain Measurements
7.11.3 Phase Identification and Structure Refinement
7.11.4 Chemical Order–Disorder Transitions
7.11.5 Short-Range Order (SRO) and Diffuse Scattering
7.11.6 In Situ X-Ray Diffraction at Synchrotrons
7.11.7 X-Ray Diffraction Measurements on Mars
Summary
Further Reading
References
Exercises
8: Diffraction of Electrons and Neutrons
8.1 Introduction
8.2 The Atomic Scattering Factor for Electrons
8.3 Basics of Electron Diffraction from Surfaces
8.3.1 Surface Reconstruction, Surface Nets, and Their Notation
8.3.2 Reciprocal Lattice Nets and Ewald Sphere Construction in Two Dimensions
8.4 Surface Electron Diffraction Methods and Applications
8.4.1 Low-Energy Electron Diffraction (LEED)
8.4.2 Adsorption Studies on Surfaces Using LEED
8.4.3 Reflection High-Energy Electron Diffraction (RHEED)
8.4.4 RHEED Oscillations: In Situ Monitoring of Thin Film Growth
8.5 Transmission High-Energy Electron Diffraction
8.5.1 Coherent, Incoherent, Elastic, and Inelastic Scattering
8.5.2 Basics of Electron Diffraction in a Transmission Electron Microscope
8.5.3 Kinematical Theory of Electron Diffraction
8.5.4 The Column Approximation, Dynamical Diffraction, and Diffraction from Imperfect Crystals
8.6 Transmission Electron Diffraction Methods
8.6.1 Selected Area Diffraction: Ring and Spot Patterns
8.6.2 Kikuchi Lines, Maps, and Patterns
8.6.3 Convergent Beam Electron Diffraction (CBED)
8.7 Examples of Transmission Electron Diffraction of Materials
8.7.1 Indexing a Single Crystal Diffraction Pattern
8.7.2 Polycrystalline Materials and Nanoparticle Arrays
8.7.3 Orientation Relationships Between Crystals or Phases
8.7.4 Chemical Order in Materials
8.7.5 Diffraction from Long-Period Multilayers
8.7.6 Twinning
8.8 Interactions of Neutrons with Matter
8.8.1 Nuclear Interactions
8.8.2 Magnetic Interactions
8.8.3 In Situ Kinetic Studies Using Neutrons: Hydration of Cement
Summary
Further Reading
References
Untitled
9: Transmission and Analytical Electron Microscopy
9.1 Introduction
9.2 Elements and Operations of a Transmission Electron Microscope
9.2.1 Electron Sources: Thermionic, Field, and Schottky Emission
9.2.2 Electromagnetic Lenses
9.2.3 The Illumination Section
9.2.3.1 Parallel Beam Operation of a TEM
9.2.3.2 Focused Probe Formation for Illumination and Scanning
9.2.4 The Imaging Section: Objective Lens and Aperture
9.2.5 Specimen Handling and Manipulation
9.2.6 The Magnification Section
9.2.7 Imaging and Diffraction Modes
9.2.7.1 Selected Area Diffraction (SAD) and Bright Field (BF) Imaging
9.2.7.2 Dark Field (DF) Imaging
9.2.7.3 Phase Contrast Imaging and the Contrast Transfer Function
9.2.7.4 Scherzer Defocus and Resolution.
9.2.8 Scanning Transmission Mode and the Principle of Reciprocity
9.2.9 Correction of Lens Aberrations
9.2.10 Image Recording and Detection of Electrons
9.3 Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods
9.3.1 Elastic Interactions
9.3.2 Mass–Thickness Contrast
9.3.3 Diffraction Contrast
9.3.4 High-Angle Incoherent Scattering: Z-Contrast Imaging
9.3.5 High-Resolution Electron Microscopy (HREM): Phase Contrast Imaging in Practice
9.3.6 Magnetic Contrast: Lorentz Microscopy
9.3.7 Electron Holography
9.4 Analytical Electron Microscopy (AEM) and Related Spectroscopies
9.4.1 Inelastic Scattering and Spectroscopy
9.4.2 Electron Energy-Loss Spectroscopy (EELS) in a TEM
9.4.2.1 The No-Loss Region
9.4.2.2 The Low-Loss Region
9.4.2.3 The Core Loss Region
9.4.2.4 The EELS Spectrometer and Signal Detection
9.4.2.5 Microanalysis Using Inner-Shell Ionization Edges
9.4.3 Quantitative Microanalysis with Energy-Dispersive X-Ray Spectrometry
9.4.4 Microdiffraction
9.5 Select Applications of TEM
9.5.1 Electron Tomography
9.5.2 Analysis of Defects: Dislocations and Stacking Faults
9.5.3 Thin Films and Multilayers: An Example
9.5.4 TEM in Semiconductor Manufacturing: Metrology, Process Development, and Failure Analysis
9.5.5 Dynamic Measurements in a TEM
9.6 Preparation of Specimens for TEM Observations
9.6.1 Chemical and Electrochemical Polishing
9.6.2 Ion-Beam Milling
9.6.3 Ultramicrotomy and Preparation of Biological Materials
9.6.4 Preparation of Cross-Section Specimens
9.6.5 Focused Ion-Beam (FIB) Milling
Summary
Further Reading
References
Untitled
10: Scanning Electron Microscopy
10.1 Introduction
10.2 The Scanning Electron Microscope
10.2.1 The Instrument
10.2.2 The Everhart–Thornley Electron Detector
10.2.3 Beam–Solid Interactions and Signals
10.2.4 The Incident Probe Size and Spatial Resolution
10.2.5 Depth of Field
10.2.6 Noise and Contrast in Imaging
10.2.7 Elastic and Inelastic Scattering, and Beam Broadening
10.3 Image Contrast in a Scanning Electron Microscope
10.3.1 Factors Influencing Secondary Electron Emission
10.3.2 Topographical Contrast in Secondary Electron Imaging
10.3.2.1 Surface-Tilt Contrast
10.3.2.2 Shadow Contrast
10.3.2.3 The Effect of the Average Atomic Number
10.3.2.4 Number Effect
10.3.2.5 Charging Effect
10.3.2.6 External Magnetic Fields
10.3.3 Angular Dependence of Back-Scattered Electrons and Topographic Information
10.3.4 Comparison of SEM Images with Different Operating Parameters
10.4 Channeling and Electron Back-Scattered Diffraction Patterns (EBSD)
10.5 Imaging Magnetic Domains
10.5.1 Type I and Type II Magnetic Contrast
10.5.2 Scanning Electron Microscopy with Polarization Analysis (SEMPA)
10.6 Probing Sample Composition and Electronic Structure
10.6.1 Basics of X-Ray Microanalysis in an SEM
10.6.2 Cathodoluminescence
10.7 Variations of Scanning Electron Microscopy
10.7.1 Environmental Scanning Electron Microscopy (ESEM)
10.7.2 Combined Focused Ion-Beam (FIB) and Scanning Electron Microscope
10.8 Preparing Specimens for SEM
Summary
Further Reading
References
Untitled
11: Scanning Probe Microscopy
11.1 Introduction
11.2 Physics of Scanning Tunneling Microscopy (STM)
11.2.1 Elastic Tunneling Through a One-Dimensional Barrier
11.2.2 Quantum Mechanical Tunneling Model of the STM
11.3 Basic Operation of the Scanning Tunneling Microscope
11.3.1 Imaging
11.3.2 Tunneling Spectroscopy
11.3.3 Manipulation of Adsorbed Atoms on Clean Surfaces
11.4 Physics of Scanning Force Microscopy
11.4.1 Mechanical Characteristics of the Cantilever
11.4.2 Cantilever as a Force Sensor
11.4.3 Tip–Specimen Forces Encountered in an SFM
11.5 Operation of the Scanning Force Microscope
11.5.1 Static Contact Mode for Topographic Imaging
11.5.2 Lateral Force Microscopy
11.5.3 Dynamic Noncontact Modes of Atomic Force Microscopy
11.6 Scanning Force Microscopy Instrumentation
11.7 Artifacts in Scanning Probe Microscopy
11.7.1 Probe Artifacts
11.7.2 Instrument Artifacts
11.8 Select Applications of Scanning Force Microscopy
11.8.1 Atomic Fingerprinting in Frequency Modulated Atomic Force Microscopy
11.8.2 Magnetic Force Microscopy (MFM)
11.8.3 Scanning Thermal Microscopy (SThM)
11.8.4 Applications of Atomic Force Microscopy in the Life Sciences
11.8.4.1 Imaging of Biomolecules
11.8.4.2 Imaging of Lipid Membranes
11.8.4.3 Imaging of Bacterial Cells
11.8.4.4 Studies of Mammalian Cells
11.8.4.5 Biological Force Spectroscopy
11.8.5 Dip-Pen Nanolithography (DPN)
Summary
Further Reading
References
Untitled
Summary Tables
Table 12.1 Spectroscopy and Chemical Methods
Table 12.2 Diffraction and Scattering Methods
Table 12.3 Imaging Methods
Index
Table of Values
Periodic Table of the Elements

Citation preview

OUP CORRECTED – FINAL

P R I N C I P L E S O F M AT E R I A L S C H A R AC T E R I Z AT I O N A N D ME T ROL O G Y

OUP CORRECTED – FINAL

OUP CORRECTED – FINAL

Principles of Materials Characterization and Metrology Kannan M. Krishnan University of Washington, Seattle

1

OUP CORRECTED – FINAL

3

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © Kannan M. Krishnan 2021 The moral rights of the author have been asserted First Edition published in 2021 Impression: 1 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2020952840 ISBN 978–0–19–883025–2 (hbk.) ISBN 978–0–19–883026–9 (pbk.) DOI: 10.1093/oso/ 9780198830252.001.0001 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

OUP CORRECTED – FINAL

To Amma, Appa, MN, and my students—past, present, and future

OUP CORRECTED – FINAL

OUP CORRECTED – FINAL

Preface

Materials science and engineering (MSE) is a multidisciplinary field, impacting every aspect of our technological society today. At the heart of MSE is understanding the relationship between structure and properties of materials. In fact, it is now well established that, by optimizing composition and structure ranging from the macroscopic to atomic dimensions, the properties of materials can be not only well controlled but also tailored for any specific application. In this endeavor, materials characterization and analysis involving a range of diffraction, imaging, and spectroscopy methods, at relevant length scales, has enabled the structure– property–processing–performance tetrahedron that epitomizes the field. Traditionally, an undergraduate curriculum in MSE emphasizes the practical application of optical microscopy and spectroscopy, imparts a working knowledge of X-ray diffraction and, where resources are available, scanning and transmission electron microscopy, and atomic force microscopy. However, recent advances in developing materials for a wide range of applications, emphasizing atomic-scale tailoring of microstructure and exploiting size-dependent properties, require an interdisciplinary approach to materials development where a judicious use of available characterization methods becomes important. This requires a coherent discussion of the underlying physical principles of materials characterization and metrology using the wide range of electrons, photons, ions, neutrons, and scanning probes. Following a broad introduction (§1), this book lays the foundations of characterization, analysis, and metrology and builds on concepts that should be familiar to an upper-division student in any branch of science or engineering. Starting with atomic structure, we develop spectroscopy methods based on intra-atomic electronic transitions (§2), followed by bonding and the electronic structure of molecules and solids motivating a number of spectroscopy methods (§3). We then discuss the periodic arrangement of atoms and develop principles of crystallography (§4), which leads to an introduction to diffraction in both real and reciprocal space. Next, we address different probes and present relevant details of the generation and use of photons, electrons, ions, neutrons, and scanning probes (§5), followed by a presentation of ion-based scattering methods (§5). A concise introduction to optics, optical microscopy, polarization of light, and ellipsometry follows (§6). The second part of the book includes a comprehensive discussion of diffraction and imaging methods that emphasize techniques widely used in the characterization and analysis of materials. This includes X-ray (§7), electron (§8), and neutron (§8) diffraction, as well as transmission and analytical electron (§9), scanning electron (§10), and scanning probe (§11) microscopies. Throughout the text, the characterization techniques are also used to introduce

OUP CORRECTED – FINAL

viii Preface and illustrate fundamental properties and materials science concepts encountered in a wide range of materials. The book is generously illustrated throughout with figures, data tables, comparison between related methods, worked examples, and concludes (§12) with three unique and comprehensive tables summarizing the salient points of all the spectroscopy, diffraction, and imaging methods presented. To keep the overall extent of the book to a manageable length, I have mainly emphasized probe-based techniques. Other methods, such as thermal characterization and property measurements, including mechanical testing, are deliberately not included. However, I do include numerous applications of materials and structures in fields ranging from science, technology, art, and biology. Each chapter also includes a number of worked examples to help tie together the concepts introduced therein, an extensive set of test-your-knowledge questions to help readers consolidate understanding of the subject matter based on the text, problem sets to further deepen learning, a summary highlighting the key concepts and ideas presented in each chapter, and an extensive bibliography for further reading. While specialized books do exist for most of the techniques discussed here, including encyclopedias of materials characterization, a coherent textbook on materials characterization and metrology at the undergraduate or early graduate level, emphasizing fundamental physical principles, such as this one, is both lacking and highly desirable. Combining discussions of the underlying principles with practical examples, and detailed sets of exercises, this book will ideally serve as a text for a year-long course at the undergraduate and/or early graduate level. Completion of such a course should give students an entry into the interdisciplinary field of Materials Science and Engineering, and a solid foundation in characterization and analysis methods, with the ability to select and apply the appropriate technique for any characterization problem at hand. Alternatively, the book can be adapted to a semester-length course, taking a more traditional approach by devoting about five weeks each to spectroscopy (Chapters 2, 3, 9, and 10), diffraction (Chapters 4, 7, and 8), and imaging (Chapters 6, 9–11). If one is further constrained in time to a 10-week quarter or term, as in many US universities, and which is the case for the course I have been teaching (syllabus available on request) for many years at UW—where UG students take this course after prior exposure to X-ray diffraction and the electronic structure of solids—this book may be adopted by teaching, selectively, the essential concepts of Chapters 1–6 at the approximate pace of one chapter per week, and in the last four weeks, covering scanning electron (Chapter 10) and scanning probe (Chapter 11) microscopies. Assigning term paper topics for self-study, involving more specialized techniques and their applications, including transmission electron microscopy (Chapter 9), would further strengthen student learning. For those students from different disciplines other than MSE, this book provides what they will essentially need to know in materials characterization, including additional background, at an early stage of their study. Overall, this book is expected to potentially have a wide readership and academic relevance

OUP CORRECTED – FINAL

Preface ix for teaching a course on characterization, analysis, and metrology, across multiple disciplines of engineering, physics, chemistry, geology, biology, art conservation, etc. The examples in the book are selected to reinforce this breadth of disciplines. Finally, even though this textbook is tailored for the teaching of upper-division undergraduate or early-stage graduate students, it is also written for self-study by experienced researchers, including those in industry, who realize that, to deliver a program/product satisfactorily, they need to know more about the microstructure of their materials than they currently do! In writing this book, I benefitted from discussions with numerous colleagues and teachers who generously shared their knowledge in multiple disciplines with me over the last four decades. Some of them also reviewed sections of this manuscript at various stages of development. Alphabetically they include: S. D. Bader, P. Blomqvist, S. Brück, J. N. Chapman, D. E. Cox, U. Dahmen, C. J. Echer, R. Egerton, M. Farle, P. J. Fischer, E.E. Fullerton, C. Hetherington, F. Hofer, W. Grogger, R. Gronsky, R. Kilaas, C.A. Lucas, R. K. Mishra, Y. Murakami, C. Nelson, S. Paciornik, S. J. Pennycook, L. Rabenberg, P. Rez, D. Shindo, I. K. Schuller, S.G.E. te Velthuis, N. Thangaraj, S. Thevuthasan, G. Thomas, M. Varela, D. O. Welch, T. Wen, and T. Young. I also offer a special note of thanks to my former students, Eric Teeman and Ryan Hufshmid, who have independently created, for the exercises in the book, a solution manual (available from OUP to those who adopt this book for teaching a course), and the anonymous reviewer who provided a chapter-by-chapter review of the entire book. Also, I benefitted immensely from interactions with many generations of graduate students and post-doctoral fellows at both UCB and UW who, driven by their own curiosity and interests, provided me the motivation to learn and apply a wide range of characterization methods in our research. The list is too long to acknowledge them individually here, but many of their contributions are reflected in this book. Finally, over the past many years, students of my course on Principles of Materials Characterization (MSE333) at UW have used draft chapters of this book as it has evolved over time with subsequent revisions. Their constructive feedback and relentless criticisms have significantly improved the book, making it more accessible and tailored to student teaching and learning. I am deeply indebted to all of the people mentioned here; however, I am entirely responsible for any remaining omissions, errors, or mistakes, and if they are brought to my attention, I will be more than happy to address them in subsequent revisions. This book has been many years in the making and parts of it were written during multiple residencies in a number of places. I am particularly beholden to the Whitely Center, an idyllic writing retreat at Friday Harbor, the Brahm Prakash Visiting Professorship at the Indian Institute of Science, Bangalore, the JSPS Senior Fellowship at the University of Tohoku, and the Humboldt Career Research Award at the University of Duisburg-Essen, all of which provided the right atmosphere to make substantial progress in writing this book. Kannan M. Krishnan Seattle, August 2020

OUP CORRECTED – FINAL

OUP CORRECTED – FINAL

Contents

1. Introduction to Materials Characterization, Analysis, and Metrology 1.1 1.2

1.3

1.4

1.5

Microstructure, Characterization, and the Materials Engineering Tetrahedron Examples of Characterization and Analysis 1.2.1 Ni-Based Superalloys: Ultrahigh Temperature Materials for Jet Engines 1.2.2 Unraveling the Structure of Deoxyribonucleic Acid (DNA) 1.2.3 Characterizing a Picasso Painting Reveals Hidden Secrets 1.2.4 Failure Analysis: Metallurgy of the RMS Titanic 1.2.5 Beneath Our Feet: Microstructure of Rocks and Minerals 1.2.6 Ceramic Materials: Sintering and Grain Boundary Phases 1.2.7 Microstructure and the Properties of Materials: An Engineering Example Probes for Characterization and Analysis: An Overview 1.3.1 Probes and Signals 1.3.2 Probes Based on the Electromagnetic Spectrum and Their Attributes 1.3.3 Wave–Particle Duality 1.3.4 Nature and Propagation of Electromagnetic Waves 1.3.5 Interactions of Probes with Matter and Criteria for Technique Selection Methods of Characterization: Spectroscopy, Diffraction, and Imaging 1.4.1 Spectroscopy: Absorption, Emission, and Transition Processes 1.4.2 Scattering and Diffraction 1.4.3 Imaging and Microscopy 1.4.4 Digital Imaging 1.4.5 In Situ Methods across Spatial and Temporal Scales Features of Materials Used for Characterization Summary Further Reading References Exercises

1 2 7 8 10 12 13 15 17 20 22 22 22 24 27 29 34 34 37 41 48 56 57 59 59 61 63

OUP CORRECTED – FINAL

xii Contents 2. Atomic Structure and Spectra 2.1 2.2

2.3

2.4

2.5

2.6

2.7

Introduction Atomic Structure 2.2.1 Bohr–Rutherford–Sommerfeld Model 2.2.2 Quantum Mechanical Model Atomic Spectra: Transitions, Emissions, and Secondary Processes 2.3.1 Dipole Selection Rules and Allowed Transitions of Electrons in Atoms 2.3.2 Characteristic X-Ray Emissions and their Nomenclature 2.3.3 Non-Radiative Auger Electron Emission 2.3.4 Electron Photoemission X-Rays as Probes: Generation and Transmission of X-Rays 2.4.1 Laboratory Sources and Methods of X-Ray Generation 2.4.2 X-Ray Absorption and Filtering 2.4.3 Synchrotron Sources of X-Ray Radiation X-Rays as Signals: Core-Level Spectroscopy with X-Rays 2.5.1 Instrumentation for Detecting X-Rays 2.5.2 Chemical X-Ray Microanalysis Surface Analysis: Spectroscopy with Electrons 2.6.1 Instrumentation for Surface Analysis with Electron Spectroscopy 2.6.2 Auger Electron Spectroscopy 2.6.3 X-Ray Photoelectron Spectroscopy 2.6.4 Surface Compositional Analysis with AES and XPS 2.6.5 Comparison of AES and XPS Select Applications 2.7.1 XRF Analysis of Dental and Medical Specimens 2.7.2 Environmental Science: Contamination in Ground Water Colloids Summary Further Reading References Exercises

3. Bonding and Spectra of Molecules and Solids 3.1 3.2 3.3

Introduction Bonds and Bands Interatomic Bonding in Solids 3.3.1 Ionic Bonding 3.3.2 Covalent Bonding 3.3.3 Metallic Bonding

68 69 69 69 70 78 78 85 91 97 97 98 102 105 107 107 113 119 120 123 128 134 136 137 137 138 139 140 142 143 147 148 149 152 153 154 156

OUP CORRECTED – FINAL

Contents xiii 3.4

3.5

3.6

3.7 3.8 3.9

3.10

Molecular Spectra 3.4.1 Vibrational and Rotational Modes 3.4.2 Ultraviolet and Visible Spectroscopy (UV-Vis) 3.4.3 Classical Model of Rayleigh and Raman Scattering 3.4.4 Selection Criteria for Infrared and Raman Activity Infrared Spectroscopy 3.5.1 Instrumentation for Raman and IR Spectroscopy 3.5.2 Michelson Interferometer and the Fourier Transform Infrared (FTIR) Method 3.5.3 Practice and Application of FTIR Raman Spectroscopy 3.6.1 Raman, Resonant Raman, and Fluorescence 3.6.2 Instrumentation for Raman Spectroscopy and Imaging 3.6.3 Application of Raman Spectroscopy in Chemical and Materials Analysis 3.6.4 Surface-Enhanced Raman Spectroscopy (SERS) Probing the Electronic Structure of Solids Photoemission and Inverse Photoemission from Solids Absorption Spectroscopies-Probing Unoccupied States 3.9.1 X-Ray Absorption Spectroscopy (XAS) 3.9.2 Near-Edge and Extended X-Ray Absorption Fine Structure (NEXAFS and EXAFS) Select Applications 3.10.1 Structure of Proteins Resolved by FTIR 3.10.2 Analysis of Catalytic Particles by XAS and XPS Summary Further Reading References Exercises

4. Crystallography and Diffraction 4.1

The Crystalline State 4.1.1 Lattices 4.1.2 Generalized Crystal Systems and Bravais Lattices 4.1.3 Lattice Points, Lines, Directions, and Planes 4.1.4 Zonal Equations 4.1.5 Atomic Size, Coordination, and Close Packing 4.1.6 Describing Crystal Structures—Some Examples 4.1.7 Symmetry and the International Tables for Crystallography 4.1.8 The Stereographic Projection 4.1.9 Imperfections in Crystals

158 158 162 166 169 172 173 174 177 179 179 181 182 183 185 188 197 197 200 206 206 208 208 211 212 214 220 222 223 224 227 231 233 236 239 244 247

OUP CORRECTED – FINAL

xiv Contents 4.2 4.3

4.4

The Reciprocal Lattice Diffraction 4.3.1 Bragg’s Law: Interpreting Diffraction in Real Space 4.3.2 The Ewald Construction: Interpreting Diffraction in Reciprocal Space 4.3.3 Comparison of X-Ray and Electron Diffraction Quasicrystals and the Definition of a Crystalline Material Summary Further Reading References Exercises

5. Probes: Sources and Their Interactions with Matter 5.1 5.2

5.3

5.4

Introduction Probes and Their Generation 5.2.1 Photons: Lamps and Lasers 5.2.2 Electrons: Thermionic and Field-Emission Sources 5.2.3 Neutrons 5.2.4 Ions Interactions of Probes with Matter, Including Damage 5.3.1 Photons 5.3.2 Electrons 5.3.3 Neutrons 5.3.4 Protons 5.3.5 Ions Ion-Based Characterization Methods 5.4.1 Rutherford Back-Scattering Spectroscopy (RBS) 5.4.2 Low-Energy Ion Scattering Spectroscopy (LEISS) 5.4.3 Secondary Ion Mass Spectrometry (SIMS) 5.4.4 Induction-Coupled Plasma Mass Spectrometry (ICP-MS) 5.4.5 Particle-Induced X-Ray Emission (PIXE) Summary Further Reading References Exercises

6. Optics, Optical Methods, and Microscopy 6.1 6.2

Introduction Wave Equation for Simple Harmonic Motion

250 256 257 259 262 266 267 268 269 270 277 278 278 278 282 287 288 292 293 294 303 303 304 315 315 325 328 331 336 337 338 339 339 345 346 346

OUP CORRECTED – FINAL

Contents xv

6.3 6.4 6.5 6.6

6.7 6.8

6.9

6.2.1 The Phase Angle 6.2.2 The Superposition Principle 6.2.3 Phasor Representation and the Addition of Waves 6.2.4 Complex Representation of a Simple Harmonic Wave 6.2.5 Superposition of Two Waves of the Same Frequency 6.2.6 Addition of Waves on Orthogonal Planes and Polarization Huygens’ Principle Young Double-Slit Experiment Reflection and Refraction Diffraction 6.6.1 Fraunhofer Diffraction from a Single Slit 6.6.2 Fraunhofer Diffraction from Double and Multiple Slits 6.6.3 Resolving Power of a Diffraction Grating 6.6.4 Fresnel Diffraction 6.6.5 Fresnel Half-Period Zones 6.6.6 Diffraction by a Circular Aperture or Disc 6.6.7 Zone Plates and Their Applications in X-Ray Microscopy Visually Observable: Characteristics of the Human Eye Optical Microscopy 6.8.1 Resolution: Rayleigh and Abbe Criteria 6.8.2 Geometric Optics and Aberrations 6.8.3 The Optical Microscope 6.8.4 Confocal Scanning Optical Microscopy (CSOM) 6.8.5 Metallography Ellipsometry 6.9.1 p- and s-Polarized Light Waves, and Fresnel Equations of Reflection 6.9.2 Optical Elements Used in Ellipsometry 6.9.3 Ellipsometry Measurements Summary Further Reading References Exercises

7. X-Ray Diffraction 7.1 7.2

Introduction Interaction of X-Rays with Electrons 7.2.1 Thomson Coherent Scattering 7.2.2 Compton Incoherent Scattering

347 348 350 350 351 353 356 357 359 361 363 366 369 370 371 372 373 375 376 376 378 381 386 388 392 394 397 398 400 401 402 402 408 409 409 410 413

OUP CORRECTED – FINAL

xvi Contents 7.3 7.4 7.5

7.6

7.7 7.8 7.9

7.10

7.11

Scattering by an Atom: Atomic Scattering Factor Scattering by a Crystal: Structure Factor Examples of Structure Factor Calculations 7.5.1 Face-Centered Cubic (FCC) Structure 7.5.2 Body-Centered Cubic (BCC) Structure 7.5.3 Hexagonal Close Packed (HCP) Structure 7.5.4 Cesium Chloride (CsCl) Structure Symmetry and Structure Factor 7.6.1 Crystals with Inversion Symmetry 7.6.2 Friedel Law 7.6.3 Systematic Absences The Inverse Problem of Determining Structure from Diffraction Intensities Broadening of Diffracted Beams and Reciprocal Lattice Points Methods of X-Ray Diffraction 7.9.1 The Laue Method for Single Crystals 7.9.2 Diffractometry of Powders and Single Crystals 7.9.3 Debye–Scherrer Method for Powders 7.9.4 Thin Films and Multilayers: Diffractometry, Reflectivity, and Pole Figures 7.9.5 Practical Considerations: Collimators and Monochromators Factors Influencing X-Ray Diffraction Intensities 7.10.1 Temperature Factor 7.10.2 Absorption or Transmission Factor 7.10.3 Lorentz Polarization Factor 7.10.4 Multiplicity 7.10.5 Corrected Intensities for Diffractometry and the Debye–Scherrer Camera Applications of X-Ray Diffraction 7.11.1 Measurement of Lattice Parameters 7.11.2 Crystallite or Grain Size and Lattice Strain Measurements 7.11.3 Phase Identification and Structure Refinement 7.11.4 Chemical Order–Disorder Transitions 7.11.5 Short-Range Order (SRO) and Diffuse Scattering 7.11.6 In Situ X-Ray Diffraction at Synchrotrons 7.11.7 X-Ray Diffraction Measurements on Mars Summary Further Reading References Exercises

415 422 425 425 427 428 429 431 431 431 431 432 433 437 438 439 443 445 449 451 451 453 455 457 457 459 460 460 461 464 467 468 469 471 472 473 474

OUP CORRECTED – FINAL

Contents xvii 8. Diffraction of Electrons and Neutrons 8.1 8.2 8.3

8.4

8.5

8.6

8.7

8.8

Introduction The Atomic Scattering Factor for Electrons Basics of Electron Diffraction from Surfaces 8.3.1 Surface Reconstruction, Surface Nets, and Their Notation 8.3.2 Reciprocal Lattice Nets and Ewald Sphere Construction in Two Dimensions Surface Electron Diffraction Methods and Applications 8.4.1 Low-Energy Electron Diffraction (LEED) 8.4.2 Adsorption Studies on Surfaces Using LEED 8.4.3 Reflection High-Energy Electron Diffraction (RHEED) 8.4.4 RHEED Oscillations: In Situ Monitoring of Thin Film Growth Transmission High-Energy Electron Diffraction 8.5.1 Coherent, Incoherent, Elastic, and Inelastic Scattering 8.5.2 Basics of Electron Diffraction in a Transmission Electron Microscope 8.5.3 Kinematical Theory of Electron Diffraction 8.5.4 The Column Approximation, Dynamical Diffraction, and Diffraction from Imperfect Crystals Transmission Electron Diffraction Methods 8.6.1 Selected Area Diffraction: Ring and Spot Patterns 8.6.2 Kikuchi Lines, Maps, and Patterns 8.6.3 Convergent Beam Electron Diffraction (CBED) Examples of Transmission Electron Diffraction of Materials 8.7.1 Indexing a Single Crystal Diffraction Pattern 8.7.2 Polycrystalline Materials and Nanoparticle Arrays 8.7.3 Orientation Relationships Between Crystals or Phases 8.7.4 Chemical Order in Materials 8.7.5 Diffraction from Long-Period Multilayers 8.7.6 Twinning Interactions of Neutrons with Matter 8.8.1 Nuclear Interactions 8.8.2 Magnetic Interactions 8.8.3 In Situ Kinetic Studies Using Neutrons: Hydration of Cement Summary Further Reading References Exercises

481 482 482 485 486 487 489 490 493 494 498 501 504 505 506 509 514 514 519 523 527 527 527 529 529 531 532 535 535 537 538 540 542 543 544

OUP CORRECTED – FINAL

xviii Contents 9. Transmission and Analytical Electron Microscopy 9.1 9.2

9.3

9.4

9.5

9.6

Introduction Elements and Operations of a Transmission Electron Microscope 9.2.1 Electron Sources: Thermionic, Field, and Schottky Emission 9.2.2 Electromagnetic Lenses 9.2.3 The Illumination Section 9.2.4 The Imaging Section: Objective Lens and Aperture 9.2.5 Specimen Handling and Manipulation 9.2.6 The Magnification Section 9.2.7 Imaging and Diffraction Modes 9.2.8 Scanning Transmission Mode and the Principle of Reciprocity 9.2.9 Correction of Lens Aberrations 9.2.10 Image Recording and Detection of Electrons Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods 9.3.1 Elastic Interactions 9.3.2 Mass–Thickness Contrast 9.3.3 Diffraction Contrast 9.3.4 High-Angle Incoherent Scattering: Z-Contrast Imaging 9.3.5 High-Resolution Electron Microscopy (HREM): Phase Contrast Imaging in Practice 9.3.6 Magnetic Contrast: Lorentz Microscopy 9.3.7 Electron Holography Analytical Electron Microscopy (AEM) and Related Spectroscopies 9.4.1 Inelastic Scattering and Spectroscopy 9.4.2 Electron Energy-Loss Spectroscopy (EELS) in a TEM 9.4.3 Quantitative Microanalysis with Energy-Dispersive X-Ray Spectrometry 9.4.4 Microdiffraction Select Applications of TEM 9.5.1 Electron Tomography 9.5.2 Analysis of Defects: Dislocations and Stacking Faults 9.5.3 Thin Films and Multilayers: An Example 9.5.4 TEM in Semiconductor Manufacturing: Metrology, Process Development, and Failure Analysis 9.5.5 Dynamic Measurements in a TEM Preparation of Specimens for TEM Observations 9.6.1 Chemical and Electrochemical Polishing 9.6.2 Ion-Beam Milling 9.6.3 Ultramicrotomy and Preparation of Biological Materials 9.6.4 Preparation of Cross-Section Specimens 9.6.5 Focused Ion-Beam (FIB) Milling

552 553 558 558 560 565 568 570 571 573 584 587 588 588 589 592 595 598 601 609 613 617 619 623 641 648 650 650 655 658 660 664 666 668 669 669 669 670

OUP CORRECTED – FINAL

Contents xix Summary Further Reading References Exercises 10. Scanning Electron Microscopy 10.1 10.2

10.3

10.4 10.5

10.6

10.7

10.8

Introduction The Scanning Electron Microscope 10.2.1 The Instrument 10.2.2 The Everhart–Thornley Electron Detector 10.2.3 Beam–Solid Interactions and Signals 10.2.4 The Incident Probe Size and Spatial Resolution 10.2.5 Depth of Field 10.2.6 Noise and Contrast in Imaging 10.2.7 Elastic and Inelastic Scattering, and Beam Broadening Image Contrast in a Scanning Electron Microscope 10.3.1 Factors Influencing Secondary Electron Emission 10.3.2 Topographical Contrast in Secondary Electron Imaging 10.3.3 Angular Dependence of Back-Scattered Electrons and Topographic Information 10.3.4 Comparison of SEM Images with Different Operating Parameters Channeling and Electron Back-Scattered Diffraction Patterns (EBSD) Imaging Magnetic Domains 10.5.1 Type I and Type II Magnetic Contrast 10.5.2 Scanning Electron Microscopy with Polarization Analysis (SEMPA) Probing Sample Composition and Electronic Structure 10.6.1 Basics of X-Ray Microanalysis in an SEM 10.6.2 Cathodoluminescence Variations of Scanning Electron Microscopy 10.7.1 Environmental Scanning Electron Microscopy (ESEM) 10.7.2 Combined Focused Ion-Beam (FIB) and Scanning Electron Microscope Preparing Specimens for SEM Summary Further Reading References Exercises

11. Scanning Probe Microscopy 11.1 11.2

Introduction Physics of Scanning Tunneling Microscopy (STM)

671 674 675 684 693 694 694 694 698 699 701 703 706 707 709 710 713 715 718 720 722 722 723 726 726 731 732 732 734 736 737 738 739 740 745 746 749

OUP CORRECTED – FINAL

xx Contents

11.3

11.4

11.5

11.6 11.7

11.8

11.2.1 Elastic Tunneling Through a One-Dimensional Barrier 11.2.2 Quantum Mechanical Tunneling Model of the STM Basic Operation of the Scanning Tunneling Microscope 11.3.1 Imaging 11.3.2 Tunneling Spectroscopy 11.3.3 Manipulation of Adsorbed Atoms on Clean Surfaces Physics of Scanning Force Microscopy 11.4.1 Mechanical Characteristics of the Cantilever 11.4.2 Cantilever as a Force Sensor 11.4.3 Tip–Specimen Forces Encountered in an SFM Operation of the Scanning Force Microscope 11.5.1 Static Contact Mode for Topographic Imaging 11.5.2 Lateral Force Microscopy 11.5.3 Dynamic Noncontact Modes of Atomic Force Microscopy Scanning Force Microscopy Instrumentation Artifacts in Scanning Probe Microscopy 11.7.1 Probe Artifacts 11.7.2 Instrument Artifacts Select Applications of Scanning Force Microscopy 11.8.1 Atomic Fingerprinting in Frequency Modulated Atomic Force Microscopy 11.8.2 Magnetic Force Microscopy (MFM) 11.8.3 Scanning Thermal Microscopy (SThM) 11.8.4 Applications of Atomic Force Microscopy in the Life Sciences 11.8.5 Dip-Pen Nanolithography (DPN) Summary Further Reading References Exercises

12. Summary Tables Table 12.1 Table 12.2 Table 12.3

Spectroscopy and Chemical Methods Diffraction and Scattering Methods Imaging Methods

749 750 752 753 755 757 759 760 763 765 767 769 772 773 776 777 777 780 780 781 782 785 786 793 794 795 796 799 803 804 814 818

Index

825

Table of values

847

Periodic Table of the Elements

848

OUP CORRECTED – FINAL

Introduction to Materials Characterization, Analysis, and Metrology

This illustration, by the author, based on a cartoon by John O’Brien (The New Yorker, February 25, 1991), succinctly describes the challenges in materials characterization. We are often called upon to describe the material microstructure (rabbit) based on the measured signals (hand) in diffraction, spectroscopy, or imaging methods. Needless to say, a poor understanding of the fundamental principles underlying the characterization methods generally lead to bad experimental design, hasty interpretations, and/or erroneous conclusions.

Principles of Materials Characterization and Metrology. Kannan M. Krishnan, Oxford University Press (2021). © Kannan M. Krishnan. DOI: 10.1093/oso/9780198830252.003.0001

1 1.1 Microstructure, Characterization, and the Materials Engineering Tetrahedron

2

1.2 Examples of Characterization and Analysis

7

1.3 Probes for Characterization and Analysis: An Overview

22

1.4 Methods of Characterization: Spectroscopy, Diffraction, and Imaging

34

1.5 Features of Materials Used for Characterization

57

Summary

59

Further Reading

59

References

61

Exercises

63

OUP CORRECTED – FINAL

2 Introduction to Materials Characterization and Analysis

1.1 Microstructure, Characterization, and the Materials Engineering Tetrahedron Materials science and engineering (MSE) is an enabling and multidisciplinary field, impacting nearly every aspect of society today. The reach of MSE is enormous—advanced semiconductors have stretched the limit of highperformance computers; optical fibers have dramatically increased the bandwidth and speed of intercontinental data transmission; magnetic materials in data storage have revolutionized information access, including the proliferation of the Internet; light-weight metals, polymers, and composites have transformed aircraft design and fuel efficiency; novel batteries and fuel-cell materials power things from cell phones to public buses; and increasingly, innovation in materials is at the heart of biomedicine. As the late William Baker, past president of Bell Laboratories, put it so elegantly: “everything is made of something and always will be!” In other words, MSE exemplifies the use-inspired fundamental studies of “Pasteur’s Quadrant” (Stokes, 1997). The dramatic societal impact of the work of materials scientists and engineers can be illustrated by numerous examples; a few of them, included in Figure 1.1.1, are adapted from a special report by a distinguished panel of members of the U.S. National Academy of Engineering [1]. When we engineer any material, we tailor its properties for a specific application. This requires that it perform in a predictable and reliable manner when it is fabricated in the desirable shape or form. The latter may be a bulk material, a composite, a coating, a thin film or heterostructure, a wire or rod, a nanoparticle or its dispersion in a matrix, a surface, a nanoscale structure, a lithographically patterned element or array, etc. In other words, we make materials with sizes ranging from the atomic to the macroscopic, and dimensionality ranging from zero to three. Sometimes, the critical feature of interest in the material may be deep inside; an example is the buried interface in many modern semiconductor, magnetic, or photonic devices that are designed and fabricated in the form of thin film heterostructures. Characterization and analysis of materials is central to the practice of materials sciences and engineering. The properties of all materials are determined by their structure, by which we broadly mean the composition, electronic structure, thermodynamic state/phase, and the arrangements of their internal components. The structure of materials can be described at various length scales or levels of detail. At the atomic level, it describes the bonding and organization of atoms or molecules relative to one another. At the mesoscopic level, it refers to an intermediate-length scale, between the atomic and microscopic, where material properties are different from the bulk, often determined by quantum mechanics, and dominated by surface effects. This is also the length scale of particular interest in nanoscience and nanotechnology (Owens and Poole, 2008). At

OUP CORRECTED – FINAL

Microstructure, Characterization, and the Materials Engineering Tetrahedron 3

Strength / density

8

6

Composites

4

2

Wood Stone

Bronze

Cast Iron

Steel

Aluminum

Engine operating temperature (°C)

1600

Carbon Fibers

(a)

1900

(b) Turbojet

1200

800

Steam Engine

400

1930s air-cooled aircraft engine

0 1800

0 1900

2000

1920

1940

1960

1980

Nd2Fe14B

35 30

Sm2Co17

28 24 20 RE2Co5

16 12 8 4

AlNiCo Steels

1900

1920

(d)

125 K

1.0

23K

102

104 103

17 K

}

Optical Fiber

Optical } Glass Phoenician

7

1953 Year

1973

1988

1980

1000 700 500 300

Poly. Diamond Cubic BN Ceramic Cermet

100 70 50

Sintered WC

30

105

4.2 K

Tool cutting speed

Optical loss in glass

39K

1960

(f)

(e)

10

103

1940 Year

3000

0.1 Superconductor critical temperature (K)

(c)

40

Year

Year

1911

Strength of permanent magnets, BHmax

45

10

Egyptian

10

3000BC1000AD 1900 1966 19791988

Year

High-speed Carbon tool steel steel 10 1750 1800 1850 1900 1950 2000 Year

Figure 1.1.1 The societal impact of materials science and engineering is illustrated by a few representative examples. (a) Strength-to-density ratio of structural materials has increased fiftyfold, compared to cast iron used two centuries ago, with application ranging from light-weight eye glasses to composite airplanes. (b) The design of high-efficiency engines with reduced environmental impact requires materials that are strong at high temperature—superalloys and specialty ceramics can operate at temperature as high as 1,100–1,400◦ C, with theoretical efficiencies of ∼80%. (c) The strength of a permanent magnet, given by its energy product, determines the design of smaller and more powerful motors—a 100fold increase from the 1930s is evident. (d) Progress in the critical temperature of superconducting materials. (e) Optical fibers are now 100 times more transparent than they were in the 1960s. (f) New hard abrasive materials have increased cutting tool speeds by a factor of 100 from the early twentieth century, making manufacturing processes cheaper and more efficient. Adapted from [1].

the microscopic—that which can be observed by some type of microscope— level, it refers to the arrangements of larger groups of atoms, such as grains and thermodynamic phases, including their morphology, chemistry, and crystallographic relationships. As materials scientists/engineers, when we use

OUP CORRECTED – FINAL

4 Introduction to Materials Characterization and Analysis Synthesis & Processing

Characterization Performance Structure

Properties

Figure 1.1.2 The materials engineering tetrahedron, where characterization plays a central role.

1 https://www.astm.org/studentmember/ metallurgybycommittee.html

the term microstructure, we generally mean all relevant structural details described here, from the atomic up to the microscopic-length scale. Lastly, at the macroscopic level, we refer to structure that can be viewed with the naked eye. The core activities of a materials scientist/engineer can be represented by a tetrahedron (Fig. 1.1.2). Naturally, we begin with the synthesis of any material and then process it to achieve the desired structure, which in turn, determines its properties and its required performance in an economical and socially acceptable manner. Note that characterization, or evaluating the microstructure at the appropriate length scale, plays a critical role in tailoring the synthesis and processing of materials to achieve the desired properties and performance of any engineering component. So, we conveniently place characterization in the center of this tetrahedron. Note that characterization, as discussed in this book, does not include measurements of properties—mechanical, magnetic, electrical, thermal, etc.—that can be found in other specialized textbooks. A range of characterization methods is required to elucidate the processing– structure–property–performance tetrahedron exhibited by the wide variety of materials that are tailor-made for specific functionality today. The list is long and includes metals/alloys, ceramics, polymers, amorphous materials and glasses that do not express any long-range crystallographic order, semiconductors, biological or biomimetic materials, composites, and natural materials like wood and paper. Alternatively, we can also classify materials in terms of their application, i.e. where structural, mechanical, functional (electrical, magnetic, optical), nuclear, biocompatibility properties, either individually or in some combination, become important. It may well be a soft material that is susceptible to damage when probed. Irrespective of the class of material that may be of interest, understanding the role of microstructure and tailoring it to optimize its properties is central to MSE. The microstructure of interest may be a combination of chemical, electronic, structural, crystallographic, or magnetic (domains) features, and has to be elucidated at the appropriate length scale that describes the behavior of the material. The characterization methods are many and the one to be applied for a specific problem has to be chosen judiciously among those readily available. Table 1.1.1 provides a list of probe-based characterization methods discussed in this book, especially those commonly identified by acronyms. Broadly, the methods of characterization using different probes are classified as diffraction, spectroscopy, and imaging. The physical principles underlying these methods are the foundations of this book. Various techniques follow from these principles and are presented, with varying detail, as appropriate for this comprehensive presentation; naturally, detailed discussions of individual techniques abound in more advanced texts, including encyclopedias of materials characterization (see Further Reading). For working engineers, the American Society of Testing and Materials, now known as ASTM International, provides detailed standards1 for a variety of materials characterization

OUP CORRECTED – FINAL

Microstructure, Characterization, and the Materials Engineering Tetrahedron 5 Table 1.1.1 Probe-based characterization methods, including acronyms where appropriate, discussed in this book, with sections indicated. Many of these techniques are mentioned in this chapter. Technique

Acronym/Abbreviation

Section

Analytical electron microscopy Auger electron spectroscopy Atomic force microscopy Atom probe field ion microscopy Back-scattered electron imaging Biological force spectroscopy Cathodoluminescence Confocal scanning optical microscopy Convergent beam electron diffraction Dip-pen nanolithography Energy dispersive X-ray spectrometry Electron back-scattered diffraction Electron energy-loss spectroscopy Electron holography Electron probe microanalysis Electron tomography Ellipsometry Energy filtered imaging in a TEM Environmental scanning electron microscopy Extended X-ray absorption fine structure Field ion microscopy Focused ion beam milling Fourier transform infrared spectroscopy High-angle annular dark field imaging

AEM

§9.4

AES

§2.6.2

AFM AP-FIM

§11.8 §1.4.3 §10.3.3

BFS

§11.8.4.5

CL CSOM

§10.6.2 §6.8.4

CBED

§8.6.3

DPN EDXS

§11.8.5 §2.5.1.1

EBSD

§10.4

EELS

§9.4.2

EPMA

§9.3.7 §2.5.2.2

EFTEM

§9.5.1 §6.9 §9.4.2.6

ESEM

§10.7.1

EXAFS

§3.9.2

FIM FIB FTIR

§1.4.3 §9.6.5 §3.5.2

HAADF

§9.3.4 continued

OUP CORRECTED – FINAL

6 Introduction to Materials Characterization and Analysis Table 1.1.1 Continued Technique

Acronym/Abbreviation

Section

High-resolution electron microscopy Inductively coupled plasma mass spectrometry Inductively coupled plasma optical emission spectrometry Interference contrast microscopy Inverse photoemission spectroscopy Lateral force microscopy Local electrode atom probe Lorentz microscopy Low-energy electron diffraction Low-energy ion scattering spectroscopy Magnetic force microscopy Metallography Neutron scattering Optical microscopy Particle induced X-ray emission Photoemission spectroscopy Raman spectroscopy Rayleigh scattering Reflection high-energy electron diffraction Rutherford back-scattering spectroscopy Scanning electron microscopy Scanning electron microscopy with polarization analysis Scanning force microscopy Scanning probe microscopy Scanning thermal microscopy Scanning transmission electron microscopy Scanning tunneling microscopy Secondary ion mass spectroscopy Selected area diffraction Transmission electron microscopy Ultraviolet and visible spectroscopy Ultraviolet photoelectron spectroscopy Wavelength dispersive (X-ray) spectroscopy X-ray absorption near edge structure X-ray absorption spectroscopy X-ray diffraction X-ray fluorescence spectroscopy X-ray magnetic circular dichroism X-ray photoelectron spectroscopy X-ray transmission microscopy Z-contrast imaging

HREM ICP-MS ICP-OES

§9.3.5 §5.4.4 §5.4.4

IPES LFM LEAP LEED LEISS MFM

PIXE PES

RHEED RBS SEM SEMPA SFM SPM SThM STEM STM SIMS SAD TEM UV-Vis UPS WDS XANES XAS XRD XRF XMCD XPS XTM

§6.8.3.3 §3.8 §11.5.2 §1.4.3 §9.3.6 §8.4.1 §5.4.2 §11.8.2 §6.8.6 §8.8 §6.8 §5.4.5 §3.8 §3.6 §3.4.2 §8.4.3 §5.4.1 §10 §10.5.2 §11.4 §11 §11.8.3 §9.2.8 §11.2 §5.4.3 §8.6.1 §9 §3.4.2 §3.8 §2.5.1.2 §3.9.2 §3.9.1 §7 §2.5.2.1 §2.6.5 §6.6.7 §9.3.4

OUP CORRECTED – FINAL

Examples of Characterization and Analysis 7 methods encountered in their professional practice; familiarity with them and adhering to these standard practices will help advance the career of those in industry. First and foremost, this book is not a catalog of characterization techniques, but, as already mentioned, it emphasizes the physical principles underlying each of the measurement techniques. The study of measurements, broadly known as metrology, is also of much industrial relevance with many companies having separate metrology departments to monitor the production and characterization of materials, devices, and components. The contents of this book are also relevant to failure analysis, an interdisciplinary engineering subject that is discussed at length in other specialized textbooks (Brooks and Choudhury, 2001). Finally, any reader who is convinced of the importance and breadth of materials characterization may skip this introductory chapter, but mastery of the basic concepts introduced here are critical for making progress through the rest of the book.

1.2 Examples of Characterization and Analysis Materials scientists and engineers are called upon to work with and characterize a wide range of materials (Callister and Rethwisch, 2009). These include metals/alloys, ceramics, polymers, semiconductors, composites, amorphous or glassy materials, wood, paper, etc. An alternative way to classify materials of interest is in terms of their application or function; for example, these include structural or mechanical properties, functional (electronic, optical, magnetic) behavior, biomaterials optimized for specific in vivo applications that raise additional issues of biocompatibility and toxicity, “smart” materials that are able to sense changes in their environment and respond to them in a predetermined manner, nanomaterials, etc. In general, the microstructural questions to be resolved are particular to the morphology of the material. For a bulk material, we may be interested in identifying its crystal structure (point or space group, Bravais lattice, unit cell dimensions, distribution of atoms in the unit cell; all described in §4), composition, chemical homogeneity, grain size and distribution, and if multi-phase, the orientation relationship between the matrix phases and the distribution of secondary phases, if any, especially at grain boundaries. We may also be interested in identifying the nature, extent, and distribution of defects (§4.1.9), be they planar, line, or point, and their effect on the properties of interest. In addition, in thin films, multilayers, and superlattices, we may wish to investigate the crystallographic relationships between the layers, i.e. texture and epitaxy.2 Questions about the nature of the

2 The nature or artificial growth of crystalline materials on single crystal substrates determining their orientation.

OUP CORRECTED – FINAL

8 Introduction to Materials Characterization and Analysis interfaces, especially the buried ones, such as roughness, compositional mixing, and in the case of magnetic materials, the spin lattice at the interface, may also be of interest. Further, elucidation of their growth mode—e.g. columnar, layerby-layer—could also be relevant (§8.4.4). For surfaces, additional questions of the surface electronic structure and changes in crystallography due to surface reconstruction also arise (§8.3.1). For zero-dimensional objects like nanoparticles, resolving their crystallography and defects, compositional homogeneity, phase purity, size, size-distribution, and shape, especially when they are on the nanoscale, is an ongoing challenge (Fig. 9.3.14). Last, but not least, for glassy or amorphous materials without long-range order, identifying a good descriptor or a method to quantify such structures is an enduring question. It is really impossible to find a single technique or an Eierlegende Wollmilchsau,3 to address these wide range of microstructural questions. To be effective, we have to identify the best technique, or set of techniques, to resolve the microstructural questions at hand. Sometimes, the optimal technique may not be readily available and one may have to make do with a less-effective alternative. Here, we begin with some typical examples of characterization and analysis of materials used in a variety of fields including engineering, biology, art, and geology. Each of these examples requires a judicious selection of characterization methods. Then, in subsequent chapters, we elucidate the fundamental principles behind these methods of spectroscopy, diffraction, and imaging, building on elementary concepts that should be familiar to the reader, i.e. atomic structure (§2), then moving to molecules (§3) and their vibration modes, and finally to crystalline solids (§4).

1.2.1 Ni-Based Superalloys: Ultrahigh Temperature Materials for Jet Engines

3 This is an old term in German for an imaginary animal, which provides everything for every purpose. A literal translation is oviparous-wooly-milking-sow. Here it means an animal that can lay eggs, give milk, and provides wool and meat. Needless to say, such an animal does not exist.

The efficiency and performance of jet engines are strongly dependent on the highest temperature attainable in their high-pressure turbine section (Fig. 1.2.1). Higher thrust requires higher operating temperatures, and for higher efficiency the engine must be made lighter without loss of thrust. Even though Ni-based superalloys in single crystal form have the required properties—high melting point (∼1,650◦ C), good thermal conductivity, low density, and intrinsic corrosion resistance—their microstructure and the ensuing thermo-mechanical properties depend on the alloying elements, their concentrations, and their processing conditions. Ni-based superalloys usually have a dual-phase microstructure (Fig. 1.2.2c) consisting of a L12 ordered γ  phase, existing in cuboidal shapes with {100} faces, separated by narrow channels of the FCC γ phase in between, and creating a coherent γ |γ  interface (Fig. 1.2.3c). The chemically ordered L12 structure

OUP CORRECTED – FINAL

Examples of Characterization and Analysis 9 HighLowCombustor pressure pressure turbine turbine (LPT) Compressors (HPT) Fan

HPT

Temperature Pressure

Figure 1.2.1 Illustration of the various components of a GE 90-115B jet engine. The variation of temperature and pressure from the back to the front of the engine is shown below. Adapted from [34].

(Fig. 1.2.3j) of the γ  phase renders it highly rigid with low dislocation tolerance; hence the dislocations are confined to the γ channels, providing the required strength and high temperature creep properties. However, when subject to thermo-mechanical loads, the microstructure evolves—forming dislocation networks, coarsening the γ  cuboids (Fig. 1.2.2d–f), and precipitating topologically close-packed phases—and deteriorates the creep resistance of the alloy. Various alloying elements and heat treatments are used to control the microstructure and its evolution upon thermo-mechanical loading: precipitation elements (Al, Ta, Ti, Re) stabilize the γ  phase, solid-solution elements (Cr, Co, Mo) strengthen the γ phase by increasing the solidus temperature and resistance to dislocation movement, grain boundary elements (C, B) form carbides and borides along the grain boundaries to prevent casting pores and strengthen low-angle grain boundaries, and oxidation resistance elements (Al) form a protective Al2 O3 surface layer. Figure 1.2.3 shows a typical microstructure analysis carried out on a single crystal superalloy after heat treatment; in practice, this is correlated with mechanical behavior like creep as described in [3]. However, in the context of characterization it is important to point out that the analysis outlined here

OUP CORRECTED – FINAL

10 Introduction to Materials Characterization and Analysis (a)

(d)

(b)

1 μm

1 μm

200 μm

200 μm

Alloy 1444+2

1 μm

Alloy 1444+4

1 μm

Eutecic

γ

(c)

(f)

γ' 500 nm

Alloy 1444+6

1 μm

Figure 1.2.2 Microstructure of Ni-based single crystal superalloys (a) in the as-cast state, and (b) after the formation of the γ  phase following heat treatment, with the morphology of the two phases shown in greater detail (c). The rate of coarsening of the γ  phase, when the samples are aged for 2 hours at 1,065◦ C, is slowed down by the addition of Re, as shown in (d) 2% Re, (e) 4% Re, and (f) 6% Re. These images were obtained using transmission electron microscopy (TEM), a technique discussed further in §9. Adapted from [2].

emphasizes the use of electron-based imaging, diffraction (§8), and spectroscopy methods in both transmission (§9) and scanning (§10) geometries.

1.2.2 Unraveling the Structure of Deoxyribonucleic Acid (DNA) “It has not escaped our notice that the specific pairing we have postulated immediately suggests a possible copying material for this genetic material”. With this characteristic understatement, the first paper [6] describing the structure of the genetic building blocks of life was published. The model of DNA proposed by

OUP CORRECTED – FINAL

Examples of Characterization and Analysis 11 (a)

(b)

(c)

(f)

γ γ 2 μm

FWHM 0.29 μm

Frequency (%)

15

5 0 0.00 0.25 0.50 0.75 1.00 γ  size (μm)

100 μm (h)

10

(e) Mean: 0.53 μm

200 nm

(g)

FWHM 0.30 μm

15 Frequency (%)

(d) Mean: 0.48 μm

2 μm

10 5

0 0.00 0.25 0.50 0.75 1.00 γ  size (μm)

(i)

(j)

200 nm

A (Al, Ta, Nb, Co) B (Ni, Co, Mo)

γ

γ

γ

γ

γ 1 nm

2 nm

Figure 1.2.3 Typical microstructure analysis of a superalloy after heat treatment showing (a) secondary electron image of the microstructure seen from the transverse directions, identifying dendritic (green, b) and interdendritic (red, c) regions, where the γ  precipitate size distribution was determined, as shown in (d) and (e), respectively. The bright field TEM image of the alloy in different regions is shown in (f ) and (g). (h) A high-resolution TEM, and (i) a high-angle annular dark field scanning TEM, HAADF-STEM image of the γ | γ  interface illustrating the high level of coherency. (j) Schematic of the L12 ordered unit cell of the γ  phase. These methods are discussed in later chapters of electron diffraction (§8), TEM (§9), and SEM (§10). Adapted from [3].

Watson and Crick (Fig. 1.2.4a) shows two phosphate-sugar, right-handed, helical chains, each coiled around the same axis, with the horizontal rods indicating the pairs of bases that holds the chains together. In particular, the bases are on the inside of the helix and the phosphates on the outside.

OUP CORRECTED – FINAL

12 Introduction to Materials Characterization and Analysis (a)

(b)

Figure 1.2.4 (a) The Watson–Crick model of DNA proposed in 1953. (b) The crucial X-ray diffraction (XRD, §7) photograph that was key to identifying the helical structure of DNA. Adapted from [7].

The crucial experiment that provided key evidence for the correctness of the Watson–Crick4 model of DNA was the X-ray diffraction (XRD) photograph (Fig. 1.2.4b) published in the same issue of the journal [7]. The specimen is a fiber with high water content and containing many millions of DNA strands aligned along the fiber axis; supposedly, this is the form of DNA in living cells. The X-ray beam is incident normal to the fiber and the X-ray photograph shows, in a striking manner, the characteristic features of helical structures. The key features of the diffraction pattern are four diamond-shaped outlines of fuzzy diffraction halos and separated by two arms of spots radiating from the center. The two arms are characteristic of helical structures, and the angle between the arms is proportional to the ratio of the width of the molecule (20 Å) to the repeat period (34 Å) of the helix. Further, careful study of the sequence of spots along the arms indicates an absence of the fourth spot, which confirms that there are only two intertwined helices involved in the structure.

1.2.3 Characterizing a Picasso Painting Reveals Hidden Secrets

4 F. H. C. Crick, J. D. Watson and M. H. F. Wilkins shared the 1962 Nobel Prize in Physiology in Medicine, and were cited “for their discoveries concerning the molecular structure of nucleic acids and its significance for information transfer in living material.”

Characterization and analysis play a significant role in art conservation, and are often used for authentication and to rule out forgery of paintings and sculptures when their provenance is questionable. Sometimes they reveal hidden secrets. Figure 1.2.5a shows Picasso’s La Miséreuse accroupie, painted early in his Blue Period era of work. Using X-ray fluorescence (§2.5.2.1) measurements in a specialized set up (Fig. 1.2.5b) to scan such large paintings, it was discovered that the current Picasso painting was painted over a landscape painting (Fig. 1.2.8c) by another artist.

OUP CORRECTED – FINAL

Examples of Characterization and Analysis 13 (a)

(b)

(c)

Figure 1.2.5 (a) The current Picasso La Miséreuse accroupie. (b) Mounting of the painting for element-specific, spatiallyresolved X-ray fluorescence measurements. (c) The hidden landscape painting buried underneath the current painting detected by element-specific X-ray fluorescence (XRF, §2) images. Adapted from the New York Times, February 21, 2018.

In X-ray fluorescence, the incident probe of X-rays is absorbed by the various elements in the pigments, which then re-emit their characteristic X-rays (or fluoresce) at specific wavelengths (§2.5.2). This element-specific X-ray fluorescence can be locally excited and mapped spatially to obtain their twodimensional distribution in the painting. Maps of iron (Fig. 2.Frontispiece.d) representing the use of Prussian blue, which is an iron-based pigment, and chromium (Fig. 2.Frontispiece.e), which is used in yellow pigments, matches the structure of the painting as seen today. However, the distributions of cadmium (Fig. 2.Frontispiece.f ) used in multiple colored pigments, including red, yellow, and orange, and lead (Fig. 2.Frontispiece.g), used as a white pigment, clearly shows a different painting underneath.

1.2.4

Failure Analysis: Metallurgy of the RMS Titanic

In the early part of the twentieth century, passengers and mail between Europe and North America crossed the Atlantic Ocean by passenger steamship. One of the most luxurious steamships to be built for this purpose was the RMS Titanic (Fig. 1.2.6a), which, on its maiden voyage in 1912, struck an iceberg that damaged its hull and broke the ship in two. Within three hours, it sank, and more than 1,500

OUP CORRECTED – FINAL

14 Introduction to Materials Characterization and Analysis (a)

(c)

10 μm

(e)

200 (b)

(d) Impact energy (Joules)

ASM Steel

150

100

Titanic Longitudinal Titanic Transverse

50

–100

0 100 Temperature (°C)

(f)

200 5 μm

Figure 1.2.6 (a) The HMSTitanic before its maiden voyage. (b) Optical micrographs (§6) of the steel from the Titanic hull in the longitudinal (top) and transverse (bottom), showing the banding with elongated pearlite colonies and MnS precipitates. (c) An SEM micrograph of the Titanic hull plate (longitudinal section) and the ASTM standard. (d) Charpy impact energy as a function of temperature for longitudinal and transverse specimens of the Titanic hull and ASTM A36 standard. (e) An SEM micrograph of a Charpy impact fracture surface newly created at 0◦ C shows cleavage planes with ledges and MnS particles; the latter is shown magnified in (f ). Adapted from [8].

of its passengers died. An oft-cited culprit for this disaster was the quality of the steel used in its construction. A metallurgical analysis of the hull steel recovered from the wreckage provides interesting insight into its failure. The first analysis of the steel looking at its overall composition Table 1.2.1 revealed a low nitrogen content, indicating that the steel was not made by the Bessemer process, which was known to render the steel against being brittle, especially at freezing temperatures. Instead, it was made by the then-alternative

OUP CORRECTED – FINAL

Examples of Characterization and Analysis 15 Table 1.2.1 The composition (at %) of steels from the Titanic, a lock gate of the same era, and ASTM A36 steel

Titanic Hull Plate Lock gate ASTM A36

C

Mn

P

S

Si

Cu

O

N

Mn:S ratio

0.21 0.25 0.20

0.47 0.52 0.55

0.045 0.01 0.012

0.069 0.03 0.037

0.017 0.02 0.007

0.024 0.01

0.013 0.018 0.079

0.0035 0.0035 0.0032

6.8:1 17.3:1 14.9:1

open-hearth process, as suggested by the relatively high oxygen and low silicon content. In addition, it contained a higher than normal phosphorus content, a very high sulfur content, and a low manganese content; the ratio of Mn:S was 6.8:1, a very low ratio by modern standards. The overall composition suggests that this steel was prone to embrittlement, or loss of ductility, especially at the freezing conditions encountered by the ship on that fateful night. Metallographic preparation for optical microscopy, consisting of grinding, polishing, and etching with 2% Nital (§6.8.5), revealed the microstructure of the steel by optical microscopy (Fig. 1.2.6b). The steel is clearly banded in the longitudinal direction (top) with an average grain size of ∼60 μm, and the pearlite phase cannot be resolved. A scanning electron micrograph (§10) reveals the pearlite colonies, ferrite grains, small non-metallic inclusions, and MnS particles (Fig. 1.2.6c) identified by energy-dispersive X-ray spectrometry (EDXS, §2.5.1.2), elongated in the direction of banding. The Charpy impact test, performed from −55◦ C to 179◦ C (Fig. 1.2.6d) shows that the ductile–brittle transition temperature is 32◦ C for the hull steel, and −27◦ C, for the comparable ASTM A36 steel. The sea water temperature at the time of the collision was −2◦ C! Note that the ASTM A36 standard has a higher Mn:S ratio, and a substantially lower phosphorous content, both of which lead to reduced ductile-brittle transitions. A scanning electron microscope (SEM) image (Fig. 1.2.6e) shows a fractured lenticular MnS particle that protrudes edge-on from the fractured surface; further, slip lines radiating away from the MnS particle can be seen. Based on such analysis [8] it can be concluded that, even though the hull steel used was the best available in 1909–1911—-when the Titanic was built—it would not meet the standards for plate steel used in ship construction today.

1.2.5 Beneath Our Feet: Microstructure of Rocks and Minerals Common materials used in engineering are sourced from common minerals, which in turn are the major constituents of common rocks. Earth is a series of “shells”; it has a liquid core composed mainly of iron (and some nickel),

OUP CORRECTED – FINAL

16 Introduction to Materials Characterization and Analysis an intermediate mantle (solid rock rich in oxygen, silicon, iron, and magnesium, in the form of silicates), and an outer crust (which averages around 30 km in thickness) that is made mostly of aluminosilicates, alkali elements, and calcium. The most abundant chemical element in the Earth’s crust is oxygen (47 wt%, 94 at%), followed by silicon (28 wt%, 1% at%), and aluminum (8 wt%, 0.5 at%). Needless to say, most metals and ceramics are extracted from this outer crust. The most common minerals are chemical compounds of silicon, aluminum, and oxygen, with small amounts of other elements distributed in them. Silicates dominate the minerals in the crust, the most abundant (58%) being feldspars (Orthoclase—KAlSi3 O8 , Albite—NaAlSi3 O8 , Anorthite—CaAl2 Si2 O8 ), followed (13%) by pyroxenes and amphiboles (Diopside—CaMgSi2 O6 , Enstatite— MgSiO3, Tremolite), and to a lesser extent, ∼10–11% each, by quartz (SiO2) and mica (Muscovite—KAl2 (AlSi3 O10 )(OH)2 , Kaolinite—Al2 Si2 O3 (OH)4 ), respectively. Petrologists study the mineralogical and chemical details of rocks, and structural geologists study the structural aspects of minerals and rocks, especially from the viewpoint of deformation processes. Detailed studies of rocks at the microscopic level provide a link between these two areas of study (Vernon, 2004). Microstructures of standard thin or polished sections of rocks are routinely observed in optical microscopes (§6.8.3) using polarized light (Fig. 1.2.7a). Cathodoluminescence (CL, §10.6.2) is used (Fig. 1.2.7b) to reveal the internal microstructure of grains, especially from those minerals (quartz, feldspar, and calcite) that appear colorless in light microscopes. Moreover, CL arises from imperfections in the crystal lattice (§4.1.9), e.g. impurity atoms, vacancies, and dislocations that are produced during the formation and growth of the minerals. Higher resolution images can be obtained in an SEM (§10), using secondary or back-scattered electrons (Fig. 1.2.7c) the latter with element sensitivity, and are used to reveal detailed microstructures of small grains, finegrain aggregates, and intergrowths. Transmission electron microscopes (§9) can provide further details of finer features, e.g. exsolution lamellae. Computed tomography (§9.5.1) maps the variation of X-ray attenuation within a solid, along multiple directions [10]. The attenuation varies with the amount of each mineral present, and a series of cross-section or 2D images, produced along different directions, are computed to provide a 3D representation of the grains and aggregates in the rock (Fig. 1.2.7d,e). Finally, electron probe microanalysis (EPMA, §2.5.2.2) and mapping produces maps of compositional distribution, particularly in fine-grained aggregates of minerals (Fig. 1.2.7c). In general, the microstructure of rocks is a product of a complicated sequence of geological events and processes. As such, its microstructural analysis, including its chemical information, applying many of the techniques discussed in this book, can provide insight into the rock’s formation, its geological history, and mineral value.

OUP CORRECTED – FINAL

Examples of Characterization and Analysis 17 Pol

(a)

B

(c) 2

(d)

4 cm

10

1

BSE

9 6 7

Si

300 μm

(e)

CL

(b)

Al

Ca

Fe

Ti

4 cm

Figure 1.2.7 A specimen of hydrothermal quartz imaged by (a) polarizing and (b) cathodoluminescence (CL, §10) microscopy. Note that CL reveals internal structures and growth zoning that is not visible in the former. Adapted from [9]. (c) Back-scattered electron image (§10) of a specimen of peletic schist from Antarctica, along with composition maps for the principal elements. X-ray tomographic images of garnetiferous metamorphic rocks. (d) A single slice of a peletic schist—garnets are light gray to white ovals, kainite appears as medium gray laths, and dark gray to black regions are rich in quartz, feldspar, and muscovite. (e) A perspective view of a 3D density map. The single slice at the bottom of the stack locates the cutaway block in the interior of the specimen. Adapted from [10].

1.2.6

Ceramic Materials: Sintering and Grain Boundary Phases

Ceramic materials are typically heterogeneous, multiphase materials, often containing crystalline and glassy (non-crystalline) phases with unique properties that make them suitable for high-temperature structural applications. As mentioned in §1.1, microstructure that may be too small to be seen with the naked eye, plays an important factor in the final property of a material. For ceramics, the microstructure is made up of small crystals called grains (Fig. 4.1.1), and in general, the smaller the grain size, the stronger and denser is the ceramic material.

OUP CORRECTED – FINAL

18 Introduction to Materials Characterization and Analysis Typically, ceramic materials are prepared by sintering, often in the presence of additives to introduce a liquid phase during sintering to overcome the poor solid-state diffusion, and achieve high densities. For example, in the case of silicon nitride, the liquid phase is introduced by oxide additives, which form a lowtemperature eutectic5 liquid with the oxidized surface layers of the silicon nitride powder precursors. However, the glassy intergranular layer is often retained after sintering, causing a deterioration in the mechanical properties. An understanding of the intergranular layer and its structure is required to improve the performance of ceramic materials. The characterization of the microstructure, including the grain boundary phases, is carried out using multiple techniques, such as X-ray microanalysis (§2.5.2), scanning and transmission electron microscopy (§9 and §10), and X-ray (§7) and electron (§8) diffraction, which is then correlated with rigorous mechanical testing. Now, we briefly introduce the microstructural evaluation of the sintering of silicon nitride with small additions of La2 O3 , Y2 O3 , and SrO [4]. A polished specimen is first prepared and observed in an SEM (Fig. 1.2.8a) and its composition analyzed by EDXS (Fig. 1.2.8b). The polished surface reveals Si3 N4 grains (90%) with a distribution of a boundary phase at grain boundaries and multi-grain junction regions (pockets); the latter can be estimated to be ∼10% of the volume.

(a)

(b)

CPS 8.838 2600.0

5.901

4.436

3.559

2.976

2.562

2.252

2.013

A

(c)

2340.0

1.823% 100 90

A

2000.0

80

A

1820.0

70

1560.0

60

1300.0

50 40

1040.0

A 780.0

A

520.0 260.0 L• 40BSD Si3314

EHT- 25.0 KV 10.0 m

WD- 11 nm

30

A

B

MAS· ×3.00 K PHOTO=7

B

B

20

B B B B C

C B

A

C

B 10 B C C

0

0.0 10

15

20

25

30

35

40

45

50

Figure 1.2.8 (a) SEM micrograph of a polished surface of a silicon nitride specimen showing the presence of the boundary phase (white), especially in the multiple grain junctions. (b) X-ray microanalysis (§2.5.2) of the polished surface shows peaks of Si and La, with very small intensities for Y and Sr. (c) X-ray diffraction (§7) patterns showing peaks that can be indexed for β-Si3 N4 (A), La5 Si3 O12 N (B), and Y5 Si3 O12 N (C). Adapted from [4]. 5 Relating to or denoting a mixture of substances (in fixed proportions) that melts and solidifies at a single temperature that is lower than the melting points of the separate constituents or of any other mixture of them.

OUP CORRECTED – FINAL

Examples of Characterization and Analysis 19 The EDXS spectrum reveals the presence of the major sintering aid (La2 O3 ) with La peak intensities substantially more intense than that for Y and Sr. Further, routine XRD θ -2θ scans (§7.9.2, Fig. 1.2.8c) reveal the presence of a majority β-Si3 N4 and minority/boundary La5 Si3 O12 N and Y5 Si3 O12 N crystalline phases; however, it is not easy to say if the crystalline boundary phases were formed during sintering or during subsequent heat treatments. Figure 1.2.9a is a lower magnification, bright-field transmission electron microscope (TEM) image of a two-grain region that shows a near-uniform dark contrast at the grain boundary. High-resolution electron microscopy (HREM) using phase contrast (§9.2.7) reveals further details of the microstructure at the atomic level. The HREM image from the two-grain region (Fig. 1.2.9b) shows that the β-Si3 N4 grains are separated by a uniform amorphous phase ∼1.8 nm in thickness. However, a HREM image (Fig. 1.2.9c) of a three-grain junction shows evidence of a crystalline phase, with two sets of orthogonal lattice fringes (0.32 nm and 0.33 nm) present in the pocket. The fringe spacing(s) is in agreement   with the interatomic spacing for the (210) and 112 lattice planes of silicon

(c)

(d)

5 nm

(b) 5 nm

0.2 μm (a) Figure 1.2.9 (a) A lower magnification bright-field TEM image showing a straight two-grain boundary. (b) An HREM micrograph of a two-grain boundary region. Using the lattice plane spacing of β-Si3 N4 as an internal calibration, the amorphous phase thickness can be estimated to be ∼1.8 nm. (c) The multigrain junction (pocket) shows two-dimensional lattice fringes, with orthogonal spacing of 0.32 nm and 0.33 nm, respectively, corresponding to La5 Si3 O12 N. A residual glassy phase between La5 Si3 O12 N and the β-Si3 N4 grain is also found. (d) The crystallized material at the grain pocket can be indexed as the diffraction pattern of La5 Si3 O12 N. Adapted from [4].

OUP CORRECTED – FINAL

20 Introduction to Materials Characterization and Analysis lanthanum oxynitride (La5 Si3 O12 N), which is further confirmed by the selected area electron diffraction (Fig. 1.2.9d). Detailed analysis of this microstructure [4] shows that it is consistent with models [5] of sintering that predict an equilibrium thickness (∼1 nm) at two-grain boundaries. Further, the grain pockets still show the presence of an amorphous phase, suggesting that it is hard to achieve a complete recrystallization of the grain boundary phase in silicon nitride. However, the partial recrystallization of the grain boundary phase can be correlated with the superior high-temperature strength of these ceramic materials.

1.2.7 Microstructure and the Properties of Materials: An Engineering Example

6 In materials characterization, the terms specimen and sample are used interchangeably. However, in this book we will endeavor to distinguish the two. The sample is generally the overall subject of the study, and the specimen is specifically what is used or prepared to carry out a measurement or examination. For example, a chemically inhomogeneous sample may have a certain bulk/average composition; however, different specimens prepared from different parts of the sample may have different compositions because of the inhomogeneity.

We now present an engineering example that illustrates a number of microstructural features; these are of particular relevance to materials characterization. Consider a dual-phase steel alloy of iron and chromium, doped with other elements, and bulk composition Fe64 Cr25.3 Ni4.7 Mn1.34 Mo1.48 C0.04 , typically used in automobile body panels, wheels, and bumpers. A number of specimens6 of this alloy, all of the same bulk composition, were prepared and first annealed for one hour at 1,300◦ C. Then, each of them was subsequently annealed again at a different temperature, Ta , over the range, 400◦ C < Ta < 1200◦ C. Their mechanical properties were measured and the ultimate tensile strength (UTS) of the alloys varied with Ta (Fig. 1.2.10a) with a minimum at Ta = 900◦ C. Now, since all the alloys have the same chemical composition, the variation of mechanical properties can only be explained in terms of the microstructure. Figure 1.2.10b shows a typical microstructure observed in a transmission electron microscopy (§9) bright field image common to all these alloys. Clearly it has two phases with distinctly different features, i.e. dark particles distributed in a light matrix. Careful crystallographic structure analysis by transmission electron diffraction (§8) shows that the structural arrangement of the unit cells in the two phases is different—the dark particles have a face-centered cubic (FCC) structure (Fig. 4.1.12), which is referred to as the γ -austenite phase, and they are distributed in a body-centered cubic (BCC), α-ferrite matrix phase. The volume fraction, fγ , of the γ -austenite phase, is obtained by analyzing these images and it varies from 5% to 40% as a function of Ta , (Fig. 1.2.10c) with a maximum for Ta = 900◦ C. It is clear that the ultimate tensile strength depends inversely on the content, fγ , of the γ - phase. Now, it is easy to understand the mechanical behavior. Compared to the α-ferrite matrix (BCC) the γ -austenite (FCC) particles have a lower resistance to plastic deformation, a larger ductility, and lower values of the UTS. This is not the end of the story. If we carefully plot the value of UTS as a function of fγ , we see not one but two distinct curves (Fig. 1.2.10d). It turns out that these two curves correspond to different compositions of the particle and matrix phases, as confirmed by appropriate chemical analysis, using EDXS (§2.5) in a TEM

OUP CORRECTED – FINAL

Examples of Characterization and Analysis 21 800 (a)

(b)

UTS (Mpa)

750

700 γ 650

γ 10μm

600 400 500 600 700 800 800 1000 1100 1200 Annealing temperature, Ta (°C)

(d)

(c)

II

750

35 UTS (Mpa)

Volume fraction, fγ, of γ-phase (%)

40

30 25 20

I 700

650

15 10

600

5 400 500 600 700 800 800 1000 1100 1200 Annealing temperature, Ta (°C)

5

10 15 20 25 30 35 40 Volume fraction, fγ, of γ-phase (%)

Figure 1.2.10 An example of the relationship between microstructure and properties in dual-phase steel. (a) Variation of the ultimate tensile strength (UTS) with annealing temperature. (b) The two-phase microstructure of γ -austenite (FCC) precipitates in an α-ferrite (BCC) matrix. (c) The γ -phase volume fraction, fγ , as a function of annealing temperature. Note the inverse correlation with UTS. (d) The variation of UTS with fγ further depends on the composition. Adapted from Kurzydlowski and Ralph (1995).

(§9.4.3). Thus, we can see that these two phases not only differ in their mechanical properties but also their alloying content. From the perspective of this book, we can see that the microstructure involves structural (crystallography), chemical (composition), and morphological (volume fraction) aspects, among others, that need to be understood to optimize the mechanical properties of these dual-phase steel alloys.

OUP CORRECTED – FINAL

22 Introduction to Materials Characterization and Analysis

1.3 Probes for Characterization and Analysis: An Overview 1.3.1 Probes and Signals

Probe

Signal

Specimen

Sampling volume/depth

Figure 1.3.1 A general representation of materials characterization using probe and signal radiations to infer details of the microstructure of materials. In general, the probe and signal may be the same type of radiation, or they may be different.

Our general approach to materials characterization is illustrated in Figure 1.3.1. We typically use an incident radiation or probe to interrogate the material and then detect a signal that can be used to infer or interpret the microstructure of interest. The interaction of the probe with the specimen can be elastic or inelastic. By elastic, we mean that there is no difference in the energy of the probe and signal, and applies predominantly to scattering or diffraction methods. By inelastic interactions, we mean a change in energy and such dispersions of signal intensity, as a function of energy, form the basis of spectroscopy techniques. However, signals from both elastic and inelastic interactions can be spatially resolved or mapped, and such spatial intensity distributions form the basis of imaging methods. We emphasize that the probe and signal may or may not be the same; in other words, they can be different. For example, in X-ray photoelectron spectroscopy (XPS, §2.6.3), we use X-rays or photons as the incident probe, and detect the photoelectrons as the signal. In this case, it is also important to recognize that, even though the penetration depth of the incident photons or X-rays may vary, the signal will be restricted to the escape depth of the photoelectrons, which is largely limited to the surface. A counter-example would be EDXS, where the incident probe is a high-energy or “fast” electron beam in an SEM (§10) or TEM (§9), and the detected signals are the characteristic X-rays emitted from elements constituting the specimen. Alternatively, electron energy-loss spectroscopy (EELS) in a TEM (§9.4.2), which is the electron analog of X-ray absorption spectroscopy (XAS, §3.9.1), uses fast electrons as the incident probe but then detects the energy dispersion of the inelastically scattered electrons, typically in the forward direction, after passing through the specimen.

1.3.2 Probes Based on the Electromagnetic Spectrum and their Attributes The electromagnetic spectrum, extending from radio waves to γ -rays (Fig. 1.3.2) and including the relationship between energy (E ), wavelength (λ), and frequency ( f ), is a good starting point to look at various probes used in materials characterization and analysis. The visible spectrum ranging from λ ∼ 380 – 700 nm, with red (λ = 650 nm), green (λ = 530 nm), and blue (λ = 470 nm) indicated prominently, is used in optical imaging/microscopy and Raman spectroscopy. At larger wavelengths, we encounter infrared (IR) radiation used in probing vibration modes in molecules, followed by microwaves and radio waves; the latter, includes nuclear magnetic resonance (NMR) of wide use in chemistry and medical imaging. At wavelengths shorter than the visible, we find ultraviolet (UV)

OUP CORRECTED – FINAL

Probes for Characterization and Analysis: An Overview 23 Range of microstructural features Lattice spacing

Grain size Quantum Dots 10–4m

10–5m

10–6m (1μm)

100nm

10nm

10–1nm (Å) 10–2nm

1nm CuKα

Extreme UV

IR

Radio– waves 10–2eV 1012 Spins (NMR)

Hard X-rays

Visible

Microwaves

VUV

1013

1eV 1014 Vibrations

γ-rays

Soft X-rays UV

10–1eV

Wavelength

MoKα 2a0

SiL CK OK

SiK Fe CuK SEM K

10eV

100eV

1keV

10keV

1015

1016

1017

1018

Bonding

Atomic Levels

TEM Energy 100keV

1MeV

Frequency (c/s) Nuclear Levels

Figure 1.3.2 The electromagnetic spectrum illustrating the range of probes that can be used in materials characterization. In addition, we must also include scanning probes as well as ions and neutrons.

radiation, useful in probing bonding states of molecules, and further down we encounter extreme UV (EUV) radiations, followed by soft X-rays,7 hard X-rays, and γ –rays, in that order. Soft X-rays overlap in energy with the atomic levels of core electrons; for reference, the SiL (99.2 eV, λ = 12.5 nm), CK (284 eV, λ = 4.4 nm), OK (532 eV, λ = 2.3 nm), SiK (1840 eV, λ = 0.674 nm), FeK (7112 eV, λ = 0.174 nm), and CuK (8980 eV, λ = 0.138 nm) absorption edges are shown. The CuKα (8.05 keV, λ = 0.154 nm) and MoKα (17.48 keV, λ = 0.071 nm) radiations, often used as laboratory sources for X-ray diffraction, are also indicated. The radius of the n = 1 orbit in the Bohr model (§2.2) of the hydrogen atom is a0 = 0.53 Å. Twice this Bohr radius, or the diameter, 2a0 , is a critical dimension that represents the spatial extent within which most of the charge for all atoms is contained; this value 2a0 = 1.06 Å, is also indicated. However, in atoms with multiple (Z) electrons, the inner shells have radii of the order of a0 /Z because they are not shielded by the outer electrons and experience the full Coulombic interactions of the nuclear charge (+Ze). Most commercial SEMs (§10.2) and TEMs (§9.2) operate in the range of 1–30 kV and 100–300 kV, respectively. The energy ranges of these “fast” electrons straddle those of γ -rays, and are also included in Figure 1.3.2. However, we

7 Wilhelm C. Röntgen (1845–1923) was the German physicist who discovered X-rays, also known as Röntgen rays, in 1895. He received the first Nobel prize in Physics in 1901 “in recognition of the extraordinary services he has rendered by the discovery of the remarkable rays subsequently named after him.”

OUP CORRECTED – FINAL

24 Introduction to Materials Characterization and Analysis should not forget that electrons are charged particles and that Coulomb forces are very strong. Microstructural features of interest in materials characterization range from the interatomic lattice spacing (∼1 Å), to intergranular phases (∼1 nm), to grains and precipitates (10 nm ∼ 1 μm). Representative microstructural examples (§1.2) of γ and γ  phases in bulk Ni-based super alloys, semiconductor quantum dots, and intergranular amorphous phase in a sintered silicon nitride ceramic, illustrate the spatial range of materials encountered in characterization and are included for comparison in Figure 1.3.2. Note that in Figure 1.3.2, energy is not expressed in the standard SI or MKS unit of joule (J), but in unit of electron volts (eV), which is the kinetic energy gained by an electron accelerated from rest through a potential difference of 1 V. It is generally recognized that the unit of joule is so large that it is inconvenient for use in routine materials characterization. Using the electron charge (e− = −1.602 × 10−19 Coulomb), and the fact that a joule is equal to a Coulomb-Volt, we can readily show that 1 eV = 1.602 × 10−19 J. Commonly, multiples of the eV, i.e. keV (103 eV) and MeV (106 eV), are also used. Further, based on the dispersion relation in vacuum, c = f λ, where c = 299,792,458 m/s, is the velocity of light, f is the frequency, and λ is the wavelength, one can write the energy, E, of the photon8 as E = hf =

12.4 (keV) hc =   λ λ Å

(1.3.1)

where Planck’s9 constant, h = 4.136 × 10−15 eV s = 6.626 × 10−34 Js, and λ is in units of Å(= 10−10 m).

1.3.3 Wave–Particle Duality

8 The word photon was introduced by G. N. Lewis in Nature, December 18, 1926. 9 Max Planck (1858–1947) was a German physicist who was awarded the Nobel prize in Physics in 1918 “in recognition of the services he rendered to the advancement of Physics by his discovery of energy quanta.”

We have described probes within the electromagnetic spectrum in terms of their unique attributes of energy, wavelength, and frequency. In materials characterization we also tend to view the probe and signal radiations as discrete particles, e.g. electrons, photons, ions, and neutrons. The wave–particle duality— a concept central to modern physics—is the key to understanding this dichotomy. Historically, the behavior of electrons and photons provided key insight into the wave and particle nature of matter. Two important classical experiments are discussed here. The photoelectric effect: When light is incident on a clean metal surface it ejects electrons (known as photoelectrons). The intensity of the light determines the number of photoelectrons emitted but the energy of the photoelectrons depends only on the frequency, f, of the light. This is impossible to reconcile with a wave description of light because it requires that the photoelectrons be emitted

OUP CORRECTED – FINAL

Probes for Characterization and Analysis: An Overview 25 with a velocity proportional to the intensity of light. Einstein10 explained the photoelectric effect by considering the incident light as a beam of particles or photons, each with a quantum of energy, h f, such that a single photon would eject an electron from the metal surface with velocity, v, given by 1 me v2 = hf −  2

(1.3.2)

where, me is the mass of the electron, h is the Planck constant, and  is the surface work function (§3.7) required to remove a photoelectron from the solid. In this then-radical theory, increasing the intensity of light increases the number of incident photons, and leads to an increase in the number of ejected electrons without changing their velocities. In contrast, diffraction of light and X-rays, the latter from the planes of atoms in crystals, arises from interference (§6.6) and confirms their wave-like nature.

Example 1.3.1: A beam of photons illuminates a metallic surface (work function = 3.45 eV) and ejects electrons with a velocity of 765 km/s. What is the wavelength of the incident photon? What part of the electromagnetic spectrum does this photon correspond to? Solution: First we convert the work function into SI units:  = 3.45 eV = 5.527 × 10−19 J Then, applying (1.3.2) and using the values of the fundamental constants, we get the frequency of the incident photon: f =

1 2 2 me v

h

+

= 1.236 × 1015 s−1

From (1.3.1), we get λ = c/f = 2.43 × 10−7 m = 243 nm, where c = 3 × 108 m/s is the velocity of light. Thus, the wavelength of the incident photon is 243 nm, and falls in the UV region of the electromagnetic spectrum (Fig. 1.3.2).

10 Albert Einstein (1879–1955) received the Nobel prize in Physics in 1921 for “his services to theoretical physics and especially for his discovery of the law of the photoelectric effect.”

OUP CORRECTED – FINAL

26 Introduction to Materials Characterization and Analysis Electron diffraction: Electrons are deflected in electric and magnetic fields consistent with a classical particle-like behavior. In fact, such deflections are used in the design of electron spectrometers (§2.6.1.2) and systems for the magnetic imaging of domains (§9.3.6). However, one can associate a wavelength, λ, and a momentum, p, where | p | = p, with the motion of the electrons as postulated by de Broglie11 (1924), in his principle of wave-particle duality: λ = h/p

(1.3.3)

where h is the Planck constant. In fact, such wave-like behavior of electrons was verified almost immediately by Davisson12 and Gerner (1925), who demonstrated the diffraction of electrons (§8) from the surface of nickel single crystals. Thus, diffraction studies of surfaces require electrons with energies of the order of 100 eV, and such surface techniques are termed low energy electron diffraction (LEED, §8.4.1) or microscopy (LEEM).13 See Example 1.3.2. However, to probe the internal crystal structure of materials, electrons with substantially higher energy (100–200 keV), e.g. in TEMs, with wavelengths in the range (0.037–0.0251 Å) are used (§9). Such high-energy electrons can also probe the structure of surfaces in reflection mode (RHEED, §8.4.3). In addition, scattering of protons (H+ ) and He+ ions of energy ∼1.0 – 2.0 MeV, are also used in materials characterization (§5.4); typically, a 2.0 MeV He+ ion has a wavelength of 10–5 nm.

Example 1.3.2: What is the kinetic energy of an electron suitable for electron diffraction of crystalline materials?

11 L. de Broglie (1892–1987) was a French physicist who was awarded the Nobel prize in Physics in 1929 and was cited for his “his discovery of the wave nature of electrons.” 12

C. J. Davisson (1881–1958) was an American physicist who won the Nobel prize in Physics in 1937 and was cited for “the experimental discovery of the diffraction of electrons by crystals.” 13

If, in addition, spin polarized (SP) electrons are used for imaging, the resulting SPLEEM microscope can be used to image surface magnetic domain structures. See Krishnan (2016), Ch. 8.

Solution: The distances between lattice planes in a crystal are of the order of 0.1 nm (1 Å) and for diffraction of electrons their wavelengths should be of comparable magnitude. Thus, for a wavelength of 0.1 nm, the velocity of the electron is, from (1.3.3), λ = hp = mhe v . Thus, v=

h 6.626 × 10−34   = 7.27 × 106 m/s =  me λ 9.109 × 10−31 0.1 × 10−9

(1.3.4)

and its kinetic energy is E=

1 me v2 = 2.41 × 10−17 J ∼ 150.4 eV 2

where 1 eV = 1.602 × 10−19 J.

(1.3.5)

OUP CORRECTED – FINAL

Probes for Characterization and Analysis: An Overview 27

1.3.4

Nature and Propagation of Electromagnetic Waves

We briefly review the laws governing the propagation of electromagnetic waves (see §6 and/or Hecht (2002) for more details). We state here, without a detailed discussion, that Maxwell14 equations, which apply to all electromagnetic phenomena, require that the electric, E, and magnetic, H, fields that form the components of any electromagnetic radiation propagate as waves in free space. Both E and H vary along the direction of propagation, with the variation being harmonic in time, and the related displacements being always along the propagation direction. Further, the waves have only transverse and no longitudinal components. Figure 1.3.3 shows such an electromagnetic wave, propagating in the z-direction, at a particular instant of time, with E confined to the xz-plane, and H confined to the yz-plane. For this wave, the plane of vibration of the electric field, E, is the xz-plane and its direction of vibration is the x-direction. Such plane waves are said to be linearly or plane polarized. For a plane-polarized radiation, with the coordinates chosen such that E is parallel to the x-axis, assuming that the variation of E with both position, z, and time, t, are sinusoidal, we can express the electric field as Ex = Ex0 sin 2π

z λ

− ft + θx

 (1.3.6)

where, Ex0 is the amplitude of the wave, λ is its wavelength, f is the frequency, and θ x is the phase angle. Naturally, the frequency and wavelength are related, i.e. c = f λ. Note that any electromagnetic radiation, such as an X-ray beam, carries energy, and the rate of flow of this energy per unit area perpendicular to the y x E H

Figure 1.3.3 The variation of E and H, for a plane polarized electromagnetic wave propagating along the positive z-direction.

E H

H

E E⊗H z

14 James Clerk Maxwell (1831–1879) was a Scottish physicist best known for his formulation of electromagnetic theory.

OUP CORRECTED – FINAL

28 Introduction to Materials Characterization and Analysis direction of propagation is given by its intensity that is proportional to the square 2 . The unit of intensity is J/m2 /s. of its amplitude, Ex0 The associated magnetic field, H, is always perpendicular to E, and in this case varies along the y-axis. Thus Hy = Hy0 sin 2π

z λ

− ft + θy

 (1.3.7)

where all terms have been defined earlier. If we are only concerned with the variations in time, at a specific location, z, of E (or, for that matter, H) we can simplify (1.3.6) for the plane wave as Ex = Ex0 sin 2π ( ft + θx )

(1.3.8)

Further, if the phase angle, θ x , is not of interest it can be set to zero. Waves with other states of polarization do exist. A simple way to visualize them is to consider the combination of two waves of the same frequency, propagating in the same direction, (say, z-axis), but with the direction of vibration of the electric vector in the x-direction for one wave and the y-direction for the other. Then the resultant polarization will depend on the combination of their amplitudes, Ex0 and Ey0 , and their phases, θ x and θ y , which is given by Ex = Ex0 sin 2π

z λ

− ft + θx

 (1.3.9a)

and Ey = Ey0 sin 2π

z λ

− ft + θy

 (1.3.9b)

Following some mathematical manipulations, at any given position along the z-axis, we can show that the locus of positions with coordinates of Ex and Ey are, in general, an ellipse (§6.2.6). The characteristics of this ellipse depend on the amplitudes, Ex0 and Ey0 , and the phase difference, θ = θx − θy , of the two waves. The minor, y , and major, x , axes of this ellipse do not lie along the original x- and y-direction, but make an angle, ψ, with the x-direction such that Ex = Ex cos ψ + Ey sin ψ Ey = −Ex sin ψ + Ey cos ψ

(1.3.10)

where Ex and Ey can be substituted from (1.3.6). This elliptically polarized wave is shown in Figure 1.3.4. Further, when Ex0 = Ey0 , the ellipse simplifies to a circle and the wave is called circularly polarized. We discuss the application of polarized light in ellipsometry in §6.9. A word on the nature of photons is now in order. Its nature can be illustrated with reference to Figure 1.3.5, which shows its electric field component, E, propagating in the z-direction. What we see is a pulse, or wave packet, moving with

OUP CORRECTED – FINAL

Probes for Characterization and Analysis: An Overview 29 y

y x

2Ey0

ψ

x

2Ex0 E, electric field

z Figure 1.3.5 A photon can be considered as a wave packet, with its electric field component, E(z), shown at any instant of time, t. Note that the entire wave packet moves with the velocity, c, of light. Adapted from Sproul (1963).

the velocity, c, of light. Even though only a small number of oscillations are shown for clarity, in reality a typical photon emitted by an atom following a transition (Fig. 2.3.1) would have millions of oscillations. The motion of this photon or wave packet, would also follow all the properties of waves (wavelength, diffraction, polarization, etc.). Both, when light is emitted or absorbed, photons are created and absorbed as single indivisible units. Its energy is hf, where f = c/λ, and h is the Planck constant. Example 1.3.3: How many photons of wavelength, λ(nm), are required to deliver 1 J of energy? Solution: The energy of each photon is E = h f = h c/λ. Thus, the number (#) of photons required to deliver 1 J of energy is # = 1/E = λ/h c = λ × 10−9 /(6.63 × 10−34 × 2.9979 × 108 ) = 5.03 × 1015 λ.

1.3.5 Interactions of Probes with Matter and Criteria for Technique Selection Any characterization technique involving a probe will interact and invariably perturb the material (§5.3). Such perturbation has the potential to cause damage. Hence, the probe should be selected carefully to give maximum information with

Figure 1.3.4 The vibrational ellipse of the electric vector, E, of an elliptically polarized wave propagating in the z-direction (out of the plane of the paper. Note that when Ex0 = Ey0 , the ellipse becomes a circle and the radiation is called circularly polarized. Also see Figure 6.2.7.

OUP CORRECTED – FINAL

30 Introduction to Materials Characterization and Analysis minimum damage. Of all the available probes, electromagnetic radiation in the visible spectrum (light, in common parlance) causes the least damage, but at the same time the information it provides is restricted to the surface, and even that, only with modest resolution. Thus, initial observations of materials should be made with optical methods (§6) using either/both elastic and inelastic scattering strategies; this includes both imaging and spectroscopic methods. Subsequently, higher-energy X-rays can be used to probe the material at greater depths to provide crystallographic (§7) and chemical information (§2.5). If imaging at higher resolution is desired, it can be accomplished by either using an SEM with higher energy (5–30 keV) electrons (§10), or using a TEM (§9) with electrons of energy >100 keV to obtain information at the highest spatial resolution, provided electron transparent thin foil specimens representative of the behavior of interest can be made (§8 and §9). Before resorting to electron-based probes, e.g. SEM and TEM, when available, scanning probe methods (§11) may also provide useful information from the specimen surface, albeit at small areal (typically, 10 μm × 10 μm) coverage. Finally, the materials can be investigated with ions (§5.4), or neutrons (§8.8), to provide a variety of complementary information. Chapter 5 discusses in detail in the various probes available (photons, electrons, ions, neutrons), the nature of their interactions—both elastic and inelastic—with the specimen, their generation, and factors to consider in their selection and use; where relevant the discussion continues in subsequent chapters dealing with specific techniques. Here, in brief, is a preview of some important points to be considered. 1.3.5.1

Penetration Depth and Mean Free Path Length

These define different distances travelled in the material being characterized by the probe radiation. Penetration depth is a measure of how deep the electromagnetic radiation can penetrate into a material; often, it is defined as the depth at which the intensity falls to 1/e (37%) of its original value. Mean free path length is the average distance traveled by a moving particle in a material between successive impacts/collisions that modify its direction or energy. For any technique, if the probe and signal radiations are not the same and have different mean free path lengths in the material, the volume analyzed (or, the sampling depth) will be determined by the radiation—either probe or signal—with the smaller mean free path length. In general, it is impossible to provide a single, comprehensive description of the penetration depth of the radiation across the entire electromagnetic spectrum. Instead, it is only possible to discuss it in terms of some specific wavelengths of importance in materials characterization. For example, in the vicinity of the visible spectrum, IR radiation is used to probe materials based on how they are absorbed (Fourier transform IR (FTIR) spectroscopy, see §3.5.2), visible light is used to examine the specimen surface (optical microscopy, §6.8), and UV radiation is used to resolve the surface electronic structure (UV photoelectron spectroscopy, UPS, §3.8).

OUP CORRECTED – FINAL

Electron mean free path (Å)

Probes for Characterization and Analysis: An Overview 31

100 50

Au

Au

Al Au

Au

Ag

Au Au Ag

10

Ag

Be

Be Ag Ni Be

5 3

Ag Mo

5

Mo

Au

Ag

10

Be P

C

Fe

Ag

Au

C

W

Be Ag Ag Mo

W

Be

50

100

500

1000

2000

Electron kinetic energy (eV)

Adapted from [35].

Higher-energy probes, particularly X-ray radiation, have a uniform and predictable behavior in all materials. The absorption of X-rays, is defined by the attenuation coefficient, μ, (see Table 2.4.1) which increases with the average atomic number of the material, and determines the depth of penetration as I = I0 exp (−μt)

Figure 1.3.6 The mean free path of electrons as a function of energy for various metals. A universal curve, with a minimum of 4.0 Å for energies in the range ∼50–100 eV, of relevance for Auger Electron Spectroscopy (AES, §2.6.2) and X-ray photoelectron spectroscopy (XPS, §3.8), is observed.

(1.3.11)

where, I 0 is the intensity of the incident X-ray probe, and I is its intensity after being transmitted through the material of thickness, t (§2.4.2). The intensities of γ -rays show the same exponential dependence on thickness as X-rays, but with their much higher energies (∼50 keV – 50 MeV) they penetrate much larger distances. For electron probes, the mean free path length varies dramatically with their energy and the (average) atomic number of the material. For low energy (∼0–2000 eV) electrons, the mean free path length is of the order of a few Å, and curiously, for all materials, satisfy a universal curve (Fig. 1.3.6) as a function of energy. This behavior, in simple terms, can be explained as follows. For the range of energies of interest, the electrons in the solid can be approximated as a free electron gas. Then, the plasma frequency, which is a function of the mean electron-electron distance, rs , determines the loss function. The inverse of the mean free path length, λ–1 , for electron propagation, is determined by rs , which, to first order, is the same for all materials. For high-energy (E ≥ 5 keV) electrons, the penetration depth, dP , behaves as dP ∝ E 1.7 /Z, where Z is the average atomic number of the material. Neutrons have ∼1,000 times the mass of electrons but do not have an electric charge. As a result, neutrons penetrate much greater distances than electrons and X-ray photons. Precise details of the penetration of neutrons in materials depend

OUP CORRECTED – FINAL

32 Introduction to Materials Characterization and Analysis on the specific atomic species; in fact, the scattering lengths of neutrons show an erratic variation in magnitude and sign with atomic number (§8.8). The interaction of high-energy ions with materials is also complex. At low (∼eV) energy they are reflected back from the surface, following simple rules of conservation of energy and momentum; at higher energies, they interact with the material, causing atomic displacements, formation of clusters and sputtering, or the removal of atoms, ions, and electrons from the specimen. In this context, it is customary to define a stopping distance, rather than a mean-free path, in the material for the ion. Further details are in §5.4. 1.3.5.2

Resolution

In the direction of the incident beam, i.e. depth resolution, the resolution is equivalent to the penetration depth already described in the previous section. However, the resolution in a direction normal to the direction of incidence, also known as the spatial resolution, depends on the diameter of the incident beam, its wavelength, and its mean free path length in the material. It also depends on the mode of imaging or signal collection. If the image is formed using an imaging system with “lenses”, utilizing either the transmitted or reflected radiation, then the spatial resolution will depend on the wavelength of the radiation, the quality of the imaging system, including the “aberrations” inherent in the lenses, and the coherence of the source. Microscopes—conventional optical, electron in diffraction or phase contrast modes, certain X-ray, and some ion—operate in this manner. To first order, the Rayleigh criterion (Fig. 1.4.10) provides a good working rule to determine the spatial resolution of these imaging modes (also see §6.8.1). The alternative method of imaging is to focus a narrow beam of radiation onto the specimen surface and again to detect the transmitted or reflected radiation. The image is formed by rastering the incident probe on the specimen surface and recording any changes in the radiation either due to elastic or inelastic interactions. Now, the spatial resolution is determined by the diameter and wavelength of the incident radiation, and the nature of the interaction, including the degree of localization of the interaction event that constitutes the signal of interest. Confocal light microscopy using lasers (§6.8.5), SEMs (§10), and ion-based optical imaging systems are examples of this mode of imaging. It is important to recognize that the spatial resolution of a technique using such focused probes of high-intensity radiation can also be limited by the potential damage the beam can cause to the material being analyzed; this is discussed briefly in the next section and in detail in §5.3. Complementing this, temporal resolution, defined as the precision of a measurement with respect to time, is a very important criterion for designing in situ and dynamic experiments for the study of growth, morphological evolution, and response of materials to various applied stimulus (§1.4.5). Further details of the achievable depth, spatial, and temporal resolution, and the factors that influence them, are discussed for specific techniques in future chapters.

OUP CORRECTED – FINAL

Probes for Characterization and Analysis: An Overview 33 1.3.5.3

Damage

Damage to the specimen can be caused by the transfer of energy and momentum from the probe (see §5.3 for a detailed discussion). For photons, the transfer of energy in the form of heat largely causes the damage. The degree and spatial extent of the damage will be determined by the penetration of the radiation in the material, and the energy and flux of the incident photon. Specifically, for photons in the IR, visible, and UV portions of the electromagnetic spectrum, their momentum is quite small. However, higher-energy photons in the X-ray range, particularly when focused, can cause significant damage if the flux density is high, such as that obtained in synchrotron sources (§2.4.3) using zone-plate “lenses” (§6.6.7). Electrons can behave both as particles and waves (§1.3.3), and when accelerated through several hundred keV in a TEM, they can cause significant damage by breaking interatomic bonds, particularly in polymeric materials (note: bondbreaking is not so common in inorganic or metallic materials). However, if the electrons are accelerated through higher voltages (∼1 MeV), such as in highvoltage electron microscopes (HVEMs), the momentum transferred to the atomic nuclei by elastic large-angle scattering is sufficient to cause significant atomic displacements even in inorganic materials and alloys (Fig. 1.3.7). Similar atomic displacements are also caused by ions, and the extent of the damage is determined by the incident ion flux. For low flux, areas of displacement damage are isolated from each other, but at high flux the displacement damage is uniformly distributed in the specimen, resulting in the formation of an amorphous region even in crystalline materials. Note that an HVEM can be a powerful tool for in situ studies of radiation processes and the kinetics of defect formation; in fact, if it can be combined with an ion beam source such as in a tandem facility,15 where if necessary, conditions similar to that experienced in a high-flux nuclear reactor can be approximated and the ensuing damage in the material can be studied at high resolution [12]. 1.3.5.4

Specimen Preparation or Requirements

These are also to be considered in probe and technique selection. While these requirements are technique specific and are described in appropriate sections throughout the book, some typical considerations are outlined here. Optical metallography (§6.8.5) often requires the polishing of surfaces, with specialized etchants (Table 6.8.1) to provide adequate contrast to resolve the features of interest. FTIR and Raman spectroscopy are quite flexible, require no vacuum, and can investigate liquids, gases, and solids. SEM requires minimal preparation, as long as the specimen is vacuum compatible, but if it is an insulator, it will require a thin (∼10 nm) coating of a conducting layer of carbon, gold, or other metal to prevent image degradation due to charging effects (§10.8). The most important requirement for carrying out TEM experiments is the ability to prepare highquality, electron transparent thin foil specimens, which are representative of the

A

0.5 μm

– –

1210

D

E D

B1 B1

C

C

1.3.7 Complex damage, showing a variety of dislocation loops, produced by 400 keV electrons in zinc irradiated to 10–3 displacements per atom (d.p.a) at room temperature. The electron beam direction was [0001]. Adapted from [11].

15 Intermediate Voltage Electron Microscopy (IVEM)—Tandem Facility, Argonne National Laboratory, USA.

OUP CORRECTED – FINAL

34 Introduction to Materials Characterization and Analysis sample microstructure. The same holds true for X-ray transmission microscopy (XTM). For atomic resolution scanning tunneling microscopy (STM, §11.3), an atomically flat specimen that is also conducting to establish a tunneling current, is required; however, scanning force microscopy (SFM), commonly known as atomic force microscopy (AFM) does not require a conducting surface, but both STM and AFM require solid surfaces that are somewhat rigid to avoid deformation during scanning. Ion scattering techniques (§5.4) require a solid specimen that is vacuum compatible.

1.4 Methods of Characterization: Spectroscopy, Diffraction, and Imaging Three basic processes underpin the foundations of most characterization methods using probes and signals. The first, spectroscopy, is generally described by considering the probe radiation as a particle and involves absorption and emission processes. The second, scattering and/or diffraction, is described by considering the radiation as a wave, and monitoring its intensity in different directions. Imaging, a third way to characterize materials, largely follows from the first two by recording the signal in a spatially resolved fashion. Images can be formed by various contrast mechanisms arising from variations in mass, chemical composition, diffraction, and phase. In scanning probe techniques (§11), contrast is obtained by mapping various forces or registering a tunneling current between the specimen and a tip as it is moved across the surface.

1.4.1 Spectroscopy: Absorption, Emission, and Transition Processes We start with a simple Bohr model of the atom (§2.2.1), represented by electrons orbiting around a nucleus of charge +Ze. In standard X-ray notation, the electron orbits are labeled as K (n = 1) , L (n = 2) , M (n = 3) . . . , where n is the principal quantum number of the electron. Further details of the electronic structure of atoms (§2) and molecules and solids (§3), as they pertain to various materials characterization methods, are discussed later. Now, consider a multi-electron atom (Fig. 1.4.1). When a primary electron (probe) with sufficient energy, E P , to overcome the binding energy, E B , of the inner-shell electron, is incident on the atom, it removes or ejects the core electron. In this process, the primary electron is scattered in some new direction with reduced energy, E P’. Note that not all primary electrons have a close encounter with one of the core electrons to result in such an ejection. The probability of such a close encounter is given by the cross-section or the probability of the occurrence of such a collision and ejection. Alternatively, the average distance that the primary electron

OUP CORRECTED – FINAL

Methods of Characterization: Spectroscopy, Diffraction, and Imaging 35 Primary Electron (EP) M +Ze

K

L

Scattered Primary Electron (EP) Secondary Electron (ES)

Figure 1.4.1 Interaction of a primary electron with the atom resulting in a core hole, a scattered primary electron, and a secondary electron.

Primary Photon (hf )

+Ze

K

L

M PhotoElectron (EPE = hf – EB ) Figure 1.4.2 An incident photon causing an inner-shell ionization and the ejection of a photoelectron.

travels between such collisions in the material is called the mean free path length (see §5.3.5.1 for related definitions). The core electron that is removed is now free and is referred to as the secondary electron. Further, the primary electron may impart some kinetic energy, E S , to the secondary electron. By conservation of energy we can easily see that the energy lost by the primary electron, EP − EP = EB + ES , is sensitive to the binding energy of the core electron, and by measuring this loss of energy accurately, the electronic structure of the atom/material can be probed. This technique, EELS, can be implemented either to probe a surface or the bulk— the latter using a TEM (§9.4). It is also possible to map the distribution of such secondary electrons as the primary electron beam is rastered along the specimen surface, and this technique forms the basis of SEM (§10). Alternatively, instead of the primary electron if a photon of sufficient energy, EP = hf , is incident, the related process of photoionization can be realized (Fig. 1.4.2). The photon can be absorbed by the atom and a photoelectron with kinetic energy, EPE = hf − EB , can be emitted. As the K-shell electrons are bound with more energy than the L-shell electrons, for the same incident photon, they will emerge with lower kinetic energy. Atomic electron binding energies are tabulated in Table 2.2.2. However, even though, to first order, binding energies of core electrons are impervious to the nature of atomic bonding, they can be perturbed by the electronic states of the outer electrons; these include electronic bonding in molecules and the valence/conduction bands in solids (see §2.6.3, §3.7). Thus,

OUP CORRECTED – FINAL

36 Introduction to Materials Characterization and Analysis studies of the energy distribution of photoelectrons for photon incidence, i.e. XPS, is a powerful technique to study chemical states in materials. Naturally, as the K-shell electrons are more tightly bound and shielded, compared to the L-shell electrons, they are less commonly probed and L-shell photoelectrons are preferred in photoemission studies (§2.6). In both ionization processes, following the ejection of the core electron, the atom is left in an excited state with an inner-shell vacancy and can rearrange its electronic configuration to minimize its energy. This is accomplished by the transition of an outer-shell electron of higher energy, aided by the strong nuclear Coulombic attractive potential, to the vacancy created by the core hole. This can be accomplished by two competing processes. 1.4.1.1

M +Ze

K

L

X-ray Photon (hf ) Figure 1.4.3 Characteristic X-ray fluorescence following the creation of a core hole. Note that the emitted photon would have the characteristics of a wave packet illustrated in Figure 1.3.5.

In this process of atomic rearrangement (Fig. 1.4.3), the transition of the electron is accompanied by the fluorescent emission of an X-ray photon with characteristic energy equal to the difference in energy between the initial and final atomic levels. As expected, these characteristic X-ray emission energies (Table 2.3.1), are element specific and their detection by energy or wavelength dispersion form the basis of elemental analysis by energy (EDXS) or wave-length dispersive X-ray spectrometry (WDS) methods. The semiconductor detectors for energy dispersion are rather compact (§2.5.1.2) and are widely implemented in the narrow confines of both SEMs and TEMs. The alternative, wavelength dispersion analysis (§2.5.1.1) requires a rather bulky crystal detector, and is implemented in dedicated instruments called electron microprobe analyzers (§2.5.2.2). Commonly referred to as microprobes, they are popularly used by geologists to analyze mineral samples (Fig. 1.2.8b). 1.4.1.2

KLL Auger Electron M +Ze

K

L KLM Auger Electron

Figure 1.4.4 Atomic relaxation process by non-radiative Auger electron emission. Two alternative scenarios, to illustrate that the Auger electron need not be emitted from the same shell, are shown.

Characteristic X-Ray Emission

Non-Radiative Auger Emission

In this competing process (Fig. 1.4.4), the atomic rearrangement is accompanied by a non-radiative emission of a second Auger electron, again with energies characteristic of the atom (Table 2.3.2). Since three atomic levels are involved, the emitted Auger electron is labeled with three capital letters—the first represents the shell where the original vacancy is created, the second represents the shell from which the vacancy is filled, and the third represents the shell from which the Auger electron is ejected. Figure 1.4.4 shows two representative Auger electrons, KLL and KLM. Which of these two competing processes is favored? Figure 1.4.5 shows X-ray fluorescence and Auger electron yields, as a function of atomic number. It can be clearly seen that, between fluorescence and non-radiative Auger emission, the former (X-rays) is preferred for high atomic number (Z) elements, but the latter (Auger electrons) is favored by low atomic number elements. In molecules, additional transition between rotational, vibrational, and electronic levels can also be probed (Fig. 3.4.2). If a monochromatic radiation of frequency, fI (usually from a laser, but in the first experiments a high-intensity mercury arc lamp was used), irradiates a specimen (Fig. 1.4.6) it can be both

OUP CORRECTED – FINAL

Methods of Characterization: Spectroscopy, Diffraction, and Imaging 37

L3-subshell Auger

0.8 K-shell Auger 0.6

0

0

20

40 60 Atomic number

L3-subshell fluorescence

80

100

elastically (no loss of energy) or inelastically (change in energy) scattered. The scattered radiation has both intensity and polarization (§1.3.4) characteristics different from the incident radiation, and both of these attributes will also depend on the direction of observation. If the frequency content of the scattered radiation is analyzed, it will show not only the original frequency, fI (referred to as Rayleigh scattering, §3.4.2), but also pairs of additional frequencies, fS = fI ± fM (referred to as Raman16 scattering), where fM is an internal frequency that depends on molecular, rotational, and electronic transitions (Fig. 3.4.2). In molecular crystals, the Raman spectrum is further dependent on the crystal symmetry, or point group (§4.3), and the intermolecular interactions, which can result in further splitting of the Raman band. Finally, understanding Raman scattering from crystalline solids requires detailed application of group theory and lattice phonon dynamics; however, in practice, a key challenge is to avoid damage (§5.7) of the specimen under irradiation by the intense laser beam. Such details are beyond the scope of this book (see Sherwood, 1972 for details); it suffices to say that for specific materials, such as diamond, the Raman modes are very good “finger prints” for the confirmation of the structure. Basic principles of Rayleigh scattering and Raman spectroscopy are introduced in §3.6.

1.4.2

Scattering and Diffraction

An electromagnetic radiation can be considered as a continuous wave or a discrete quantum of energy, called photons. Even the quanta are discrete bundles of waves 16 Sir C. V. Raman received the Nobel prize in Physics in 1930 and was cited for “his work on the scattering of light and for the discovery of the effect named after him.” 17 It is common in optical spectroscopy to specify the radiation in inverse wavelength or wavenumber, υ = 1/λ

Adapted from [13].

(a)

(b)

(c) –762 –790

K-shell fluorescence

0.2

Figure 1.4.5 Auger electron and X-ray fluorescence yields, for K- and L-shell ionization, as a function of atomic number. It is clear that lower atomic number elements favor Auger emissions. See also Example 2.3.2.

–218 –314 –459

0.4

+459 +314 +218 ν0

Fluorescence & Auger yields

1.0

Figure 1.4.6 One of the first published Raman spectra of carbon tetrachloride liquid. (a) Spectrum of the mercury arc lamp in the region of (λ = 435.9 nm, or wavenumber υ0 = 22938 cm−1 )17 used as the incident probe. (b) Rayleigh and Raman spectrum from liquid CCl4 . (c) The principal lines from the spectrum of CCl4 are indexed in wave numbers (cm–1 ) with respect to υ0 = 22938 cm−1 Adapted from [14].

OUP CORRECTED – FINAL

38 Introduction to Materials Characterization and Analysis Bright dots appear where wave crests meet and reinforce each other

Darkness results where wave troughs and crests meet and cancel each other Diffraction Pattern Wave crest

Figure 1.4.7 Interference from a diffraction grating consisting of two slits. Adapted from https:// www.nobelprize.org/prizes/chemistry/ 2011/press-release/. For further details of diffraction gratings see §6.6. In the case of crystalline materials, the slits are replaced by a periodic arrangement of atoms.

Wave trough Diffraction Grating

Incident light

(Fig. 1.3.5) and their wavelength determines their energy, and in the case of the visible spectrum, their color. When multiple quanta reach the same point in space, they either interfere constructively (become brighter) or destructively (become darker). Simply put, the results of the interference between two waves depend on their phase difference. In the extreme case, if the two waves of the same wavelength are completely in phase (phase difference = mπ, where m is 0 or an even integer) they interfere constructively (addition of amplitudes). Alternatively, if the two waves are out of phase (phase difference = mπ , where m is an odd integer), they interfere destructively (subtraction of amplitudes). Such an interference effect is illustrated in the high-school physics experiment involving the passage of monochromatic light through two slits: a diffraction grating (Fig. 1.4.7). As the light waves travel through the grating, the waves from the two slits interfere and an alternating dark and light pattern on a screen positioned behind the grating is formed. This diffraction pattern arises, in the general sense, whenever a wave motion (incident radiation) encounters an ordered array of scatterers (such as slits in the grating), which then redirects the incident radiation into relatively well-defined directions. The only requirement for such diffraction to occur is that the wavelength of the incident radiation should be of the same magnitude as the distance between the scattering centers. In crystalline materials, instead of slits, the gratings are made up of a periodic arrangement of atoms in three dimensions (see discussion at the end of §6.6.2). The repeat distance, or the period, d, in crystalline materials is of the order of 1 Å (0.1 nm), and so X-rays (Fig. 1.3.2) with similar wavelengths are ideally suited for such diffraction experiments (see Examples 1.4.1 and 1.4.2). Alternatively,

OUP CORRECTED – FINAL

Methods of Characterization: Spectroscopy, Diffraction, and Imaging 39 electrons subject to an acceleration potential (high energy, ∼100 keV, for bulk studies in a transmission electron microscope, or low energy, ∼100 eV, for surface studies), or neutrons can also be used. The periodic array of atoms in the crystal form well-defined planes that coherently scatter the incident radiation along specific directions, given by Bragg’s18 law λ = 2 d sin θ

λ

(1.4.1)

where λ is the wavelength of the incident radiation, d is the interplanar spacing, and θ is measured from the reflecting planes (Fig. 1.4.8). The periodic arrangement of atoms in a 3D crystal describe many such planes with different interplanar spacings, and hence, diffraction peaks or positive interference occurs along several directions (Fig. 1.4.9a). In contrast, amorphous materials show a broad peak (Fig. 1.4.9b) and quasicrystal (§4.4) show unusual fivefold symmetry (Fig. 1.4.9c), a rotational symmetry inconsistent with lattice translations (see §4.1). Diffraction, particularly in reciprocal space is introduced in §4, and discussed in detail, starting in §7 for X-rays and continuing in §8 for electrons and neutrons.

θ θ d

Figure 1.4.8 Diffraction from a periodic arrangement of scattering sites, such as atoms in crystalline materials.

Example 1.4.1: A XRD measurement for NaCl (Fig. 7.9.5) shows the most intense Bragg peak for dhkl = 2.82 Å at 2θ = 31.69◦ . What is the wavelength of the X-ray radiation? Solution: Applying (1.4.1), we get λ = 2∗ 2.82 Sin (15.89) = 1.54 Å, which is the Cu Kα radiation, often used for XRD in the laboratory.

(c) Intensity

(a) crystal

Intensity

(b) liquid or amorphous solid

Diffraction (Scattering) angle, 2θ 18 Sir Lawrence Bragg, at age 25, was the youngest Nobel prize (1915) winner in Physics, and was cited for “services in the analysis of crystal structure by means of X-rays.” 19 Dan Shechtman won the Nobel prize in Chemistry in 2011, and was cited for “the discovery of quasicrystals.”

Figure 1.4.9 X-ray diffraction pattern (θ – 2θ scan) as a function of angle, 2θ, in (a) crystalline, and (b) amorphous materials. (c) An electron diffraction pattern from a quasicrystalline material, originally identified by Professor Schechtman.19 Notice the unusual tenfold rotation symmetry—a rotational axis that is inconsistent with lattice translations. Such crystals are mathematically regular but do not repeat themselves periodically.

OUP CORRECTED – FINAL

40 Introduction to Materials Characterization and Analysis λ

λ Figure 1.4.10 Scattering of incident radiation by (a) point scatterer isotropically in all directions and (b) partially ordered material scattering non-isotropically.

(a)

(b)

In addition to diffraction, incident radiation can also be redirected over a wide angular range by a point scatterer, such as an individual atom (Fig. 1.4.10a), and by a partially ordered material or a rough surface (Fig. 1.4.10b). The angular distribution of such scattering process is related to the spatial periodicities of the scattering object—obtained from the Fourier transform of its charge density distribution function. Thus, a point object scatters radiation evenly in all directions. In contrast to crystalline materials, in amorphous solids or liquids the atoms are tightly packed together with a preference, statistically speaking, for a specific interatomic spacing but with a general lack of periodicity. Figure 1.4.9 shows representative X-ray scattering from periodic crystals and amorphous solids. It is worth mentioning here that if a collimated beam of monoenergetic incident particles (typically 4 He ions with MeV energy) are used to probe a solid, the ions are kinematically scattered (§5.3.5.3) following simple rules of conservation of energy and momentum both parallel and perpendicular to the direction of incidence. It can be easily shown that the energy of the ions after scattering is determined by the masses of the particle (4 He ions) and the target atoms in the material, as well as the scattering angle. For direct back-scattering, the energy ratio of the scattered to the incident ion has the lowest value and this geometry is often used in a technique called Rutherford back-scattering spectrometry (RBS, §5.4.1) to determine composition profiles, with depth, of a wide variety of materials. RBS is particularly popular in the analysis of semiconductor heterostructures and further details are discussed in §5.4. Example 1.4.2: In a low-energy electron diffraction experiment, a peak is observed at 2θ = 24.42◦ for a specimen with an interplanar spacing of 2.9 Å. What is the voltage used to accelerate the electron in this experiment? Solution: Applying Bragg’s law, (1.4.1), we solve for the wavelength of the incident electrons:   λ = 2 d sinθ = 2∗ 2.9∗ sin 12.21◦ = 1.227 Å From the de Broglie relation, (1.3.3), we have p = me v = h/λ. Thus, the velocity of the of the incident electron is v=

h = 5.93 × 106 m/s λme

OUP CORRECTED – FINAL

Methods of Characterization: Spectroscopy, Diffraction, and Imaging 41

Its kinetic energy is E=

1 me v2 = 1.60 × 10−17 J 2

Assuming that all this kinetic energy is due to the acceleration potential, V, we get V = E/e = 100 V, where e = 1.6 × 10−19 C is the charge of the electron.

1.4.3

Imaging and Microscopy

Complementing spectroscopy and diffraction/scattering, imaging is the third principal component of materials characterization. Optical microscopy/metallography (§6.8.5) was the first technique developed by Sorby20 to reveal the microstructure of metallic surfaces. From these early observations of steels by optical methods [15], it became apparent that materials not only had structure, but more importantly, that the defects in steels could be related to their properties. This gave rise to the technique of metallography, including related surface preparation (Table 6.8.1) and laid the foundations of structure–property correlations in metallurgy and materials sciences. To coordinate with the excellent characteristics of data collection and image formation of the human eye (§6.7), optical imaging is optimized for its sensitivity range (λ = 380 − 700 nm), peaking in the green (λ = 560 nm) portion of the visible spectrum. Hence, green filters are commonly used to focus optical microscopes and the viewing screens of TEMs are coated with a green phosphor. Now, the resolution, δ, of a microscope is given by the Rayleigh criterion (§6.8.1): δ=

0.6λ n sin α

(1.4.2)

where, λ is the wavelength of the radiation, n is the refractive index of the material, and 2α is the angle that a point source subtends at the lens (Fig. 1.4.11). For a given geometry, fixing 2α, the resolution can be improved by going to smaller wavelengths, λ. Figure 1.4.12 shows a typical example that illustrates the effect of wavelength on resolution. The optical micrograph (a) at 500×, of the Toluca iron meteorite is compared with an image (b) of the same meteorite surface taken with a SEM at 2000×, with the latter showing substantially greater details. Nevertheless, optical microscopy (§6.8) is a rapid and efficient technique that is the mainstay of materials characterization; further details, including the use of crossed polarizers for non-cubic materials, are discussed in §6.8. Based on the Rayleigh criterion of resolution, (1.4.2), it is indeed tempting to consider using X-rays of shorter wavelengths than the visible spectrum to improve

20

Henry Clifton Sorby, 1826–1908.

OUP CORRECTED – FINAL

42 Introduction to Materials Characterization and Analysis Image Plane

Intens

ity

Object Plane

Two Point Sources

Figure 1.4.11 The Rayleigh criterion, defined by the separation in the image of two point-sources, such that the maxima of one coincides with the first minima of the other.

α

(a) Figure 1.4.12 (a) The lamellar (L) structure in a high-Ni Taenite (T) phase of the Toluca iron meteorite observed using an optical microscope at 500x magnification. The grey region (CT1) surrounding the lamellar structure is a high-Ni ordered FeNi phase. (b) The same lamellar structure observed in an SEM at 2000×; now the details of the lamellar structure, composed of aligned K and T phases, are clearly resolved. Adapted from Williams, Pelton, and Gronsky (1991).

(b) K

CT 1

L

20 μm

resolution. However, optical lenses are based on the simple idea of refraction, i.e. the bending of radiation at the interface between materials of different refractive indices (§6.5). Unfortunately, for EUV and soft X-rays wavelengths, significant refraction cannot be accomplished within a single absorption length. As a result, real images using refraction (lenses) at X-ray wavelengths are not practical. Instead, diffraction techniques using Fresnel zone plate lenses (§6.6.7) are employed for high-resolution soft X-ray (λ = 0.4−4.4 nm) microscopy. There

OUP CORRECTED – FINAL

Methods of Characterization: Spectroscopy, Diffraction, and Imaging 43 are two versions of such microscopes: (a) the full-field soft X-ray microscope (Fig. 6.6.12), which uses the zone plate lens after the specimen to form a complete image, point by point, much like a common optical microscope, and (b) the scanning X-ray microscope, which uses the zone plate to focus a spot of X-ray radiation on the specimen, which is then scanned and the radiation transmitted through the specimen is used to construct its image, pixel by pixel. In addition to recording the absorption as a function of position, it can also record element specific signals, such as emission or fluorescence, giving chemical information as a function of position. Alternatively, using circularly polarized light (Fig. 1.3.4) and X-ray magnetic circular dichroism (XMCD) for magnetic contrast, the magnetic microstructure, or the domain structure, can also be imaged (Fig. 1.4.13). Further discussion of various magnetic imaging methods can be found in Krishnan (2016). XTM, using synchrotron radiation, is introduced in §6.6.7.

(a)

(b)

H = 0 Oe (d)

H = 625 Oe (e)

H = 0 Oe

(c)

(f)

H = 1450 Oe

H = 1250 Oe (g)

H = 2200 Oe

H = 2800 Oe

Figure 1.4.13 Magnetic images of {Fe0.34nm Gd0.4nm ]80 multilayers under magnetic fields applied perpendicular to the film as indicated. (a-c) Images taken using a full-field soft X-ray transmission microscope (XTM), at the FeL3 edge, with contrast sensitive to the out of plane magnetization. Starting with a stripe domain (a), the domains are pinched to form cylinders in the same space (b), followed by final dissipation (c) of the cylindrical domains. (d-f) Lorentz microscopy images (§9.3.6), recorded at room temperatures, in a FEI Titan microscope equipped with an aberration corrector. Transitions from stripe domains (d), to a magnetic skyrmion21 lattice (f), and subsequently to disordered skyrmions (g) are observed. Lorentz microscopy is sensitive to the in-plane magnetic induction only. Adapted from [16].

21 A magnetic skyrmion is a quasiparticle that defines the smallest perturbation to a uniform magnetic field and is visualized as a point-like region of reversed magnetization surrounded by a whirling twist of spins.

OUP CORRECTED – FINAL

44 Introduction to Materials Characterization and Analysis

22 Ernst Ruska (1906–1988) was a German physicist who received a muchbelated Nobel prize in Physics in 1982; he was cited for “his fundamental work in electron optics, and for the design of the first electron microscope.”

Attempts to improve upon the resolution of optical microscopy by using shorter wavelength radiation, such as soft X-rays, have been successful but not easily accessible as they require synchrotron radiation sources (§2.4.3) to generate the X-ray radiation. For routine laboratory use, a more successful optical approach is to reduce the size and intensity of the light source to a sub-micron scale, i.e. using a laser, to generate image signals from individual microscopic spots on the specimen surface. Further, apertures are used to eliminate all light in the image from any plane in the specimen outside the plane of focus. In other words, this method improves resolution and contrast by employing spatial filtering methods to eliminate scattered or reflected light from planes that are out of focus. Twodimensional images are formed by rastering the spot on the specimen surface; 3D images can be constructed by changing the image plane in the vertical direction. Such confocal scanning optical microscopes (CSOM), described in §6.8.5, have found much use in imaging biological materials (Sheppard and Shotton, 1997). Following (1.4.2), electrons accelerated at higher voltages (∼100 kV), are a good alternative to achieving much smaller wavelengths, and thus superior resolutions. However, the electromagnetic lenses required for electron microscopy suffer from various lens aberrations (§9.2.2) that limit the collection angle, α, of electrons to 0.05–0.5◦ (10–2 –10–3 rad). In vacuum, the refractive index, μ = 1, and the Rayleigh criterion for small angles α, gives a resolution δ = 1.2λ/μ sin α = 1.2λ/α. Interatomic distances in solids are of the order of 0.1 nm (see §5.3.5.1 for a discussion of sizes and dimensions); thus, atomic resolution should be attainable for electrons microscopes at ∼100 kV. Further, taking into consideration the aberrations of electromagnetic lenses, and being mindful of the need to penetrate electron transparent thin foils representative of the bulk material being investigated, atomic resolution was typically achieved in microscopes operating only at higher voltages. However, recent developments in the design and availability of aberration corrected microscopes have made sub-Å resolution in transmission electron microscopy routinely achievable. Figure 1.4.14 provides a very convincing case of the advancement made in image resolution, with successive generations of high-resolution TEMs, using an example of the structural determination of β-SiAlON. From the first demonstration of a transmission electron microscope in 1933 by Ernst Ruska,22 electron microscopy and diffraction have undergone spectacular development over the years. Further details of this versatile technique, where a single instrument incorporating multiple detectors can provide comprehensive information on the physical, chemical, and magnetic microstructure (Fig. 1.4.13) of the specimen, all at unmatched spatial resolution (Fig. 9.1.2) through diffraction, imaging, and spectroscopy, are discussed in §8 and §9. Complementing TEM, the SEM (§10), in which the resolution depends on the size of a finely focused probe incident on the specimen, is a versatile instrument

OUP CORRECTED – FINAL

Methods of Characterization: Spectroscopy, Diffraction, and Imaging 45

(a)

(d)

0.76 nm

(g)

(b)

(c)

(e)

(f)

(h)

(i)

Figure 1.4.14 High resolution atomic structure images of a ceramic material, β SiAlON, taken with three generations of transmission electron microscopes at the National Center for Electron Microscopy, Berkeley. Top: The Atomic Resolution Microscope (ARM) designed to achieve atomic resolution using high voltage (1 MeV, λ = 0.00087 nm) and a spherical aberration coefficient, Cs = 2 mm. Middle: The next generation One Ångstrom Microscope (OAM), uses a 300 kV, Schottky emission gun source and a much smaller Cs = 0.6 mm. Bottom: The most recent Transmission Electron Aberration-corrected Microscope (TEAM) uses hardware corrections of the aberrations allowing Cs to be tuned at will, and designed for a resolution limit of 0.05 nm. For each microscope, the left column shows the image recorded at Scherzer defocus (§9.2.7.4), with the specimen viewed along the (0001) zone axis of this hexagonal material. The middle column shows a simulation of the structure and the right column shows an overlay of the simulated image on the atomic structure. While all three images (a, d, g) show sixfold symmetry, only the TEAM image shows a direct relationship to the atomic structure. TEM is described in detail in §9. Adapted from [17].

OUP CORRECTED – FINAL

46 Introduction to Materials Characterization and Analysis providing various modes for imaging using secondary or back-scattered electrons. It provides topographic (Fig. 1.4.12b), compositional (Fig. 1.4.15), voltage, and magnetic contrast as well as crystallographic information through channeling patterns (§10.4). Another elegant way to obtain microstructural information at atomic resolution (Fig. 1.4.16) is to apply a large electric field to a specimen of the material, shaped in the form of a needle, such that the potential barrier, V, for the electrons to leave the specimen can be locally overcome at the tip. Such field emission (§5.2.2) was used to image local variations in the work function. Alternatively, by admitting a small quantity of gas into the chamber and reversing the voltage on the needle (in the first experiments, the material happened to be tungsten) the gas close to the needle surface can be ionized. As the electric field is increased, these ionized gas atoms are repelled from the tip surface towards a fluorescent screen, producing a magnified image of the needle specimen (Fig. 1.4.16) that can clearly delineate the individual atoms. This technique, called field ion microscopy (FIM) [18, 19], has made significant contributions to the study of the structure of interfaces and grain boundaries on the atomic scale [20], but will not be discussed further in this book. A serious limitation of the FIM has been its inability to identify the chemical

(a)

(b)

(c)

10 μm

Figure 1.4.15 SEM compositional maps, using characteristic X-rays, of a Cu–Zn specimen showing diffusion-induced grain boundary migration. (a) Znmap, with Zn in white, (b) Cu-map, with 90% Cu being black, and (c) a pseudocolor superposition image with Cu in green and Zn in yellow. Adapted from Gronsky, Pelton, and Williams (1991).

(a)

Microchannel Plate Imaging Gas Atom

Figure 1.4.16 (a) Schematic of an atom probe field ion microscope (APFIM) and (b) FIM image of a Tungsten tip. Note that individual atoms are resolved. FIM was invented by E. R. Müller.

Gas Ion FIM Tip Probe Hole

Phosphor Screen Microchannel Ionized Plate Detector Atom Phosphor Screen

(b)

OUP CORRECTED – FINAL

Methods of Characterization: Spectroscopy, Diffraction, and Imaging 47 nature of the individually imaged atoms. To overcome this limitation, Müller23 conceived and built the atom-probe FIM (Fig. 1.4.16a), which is a combination of a probe-hole FIM with a mass spectrometer (§5.4.3) having single particle sensitivity [21]. During observation, the observer selects an atomic site of interest by placing it over a probe hole in the image screen. Pulsed field evaporation sends the chosen particle through the hole and into the spectrometer section. The APFIM (see Miller et al., 1996 for a detailed discussion) has undergone numerous major improvements, including the imaging atom probe [22] and the positionsensitive atom probe [23]. The most common geometry currently used is a local electrode atom probe (LEAP) [24], which first appeared commercially in 2002. The current status of atom probe tomography is discussed in [19]. If such a “needle” reverses its role, i.e. it is used as a probe, rather than as the specimen, and is mounted on a flexible cantilever, which is then brought very close—within a few atomic distances—to the specimen surface, a tunneling current can be established between the specimen and the tip. If the tip is now scanned, the specimen surface in vacuum can be imaged at atomic resolution, either by monitoring the tunneling current at constant height, or by monitoring the height at constant tunneling current (§11.2). Such an instrument is called an STM and was invented by Binnig and Rohrer.24 Moreover, the sharp tip can serve as a fine probe to measure a variety of physical properties on the local scale by using a variety of spectroscopic methods. Note that the STM resolves individual atoms on conducting surfaces (Fig. 1.4.17). For resolving individual atoms on insulating surfaces the alternative atomic force microscope (AFM) was introduced [26]. Here, instead of the tunneling current, atomic forces between the needle and the specimen surface, as the tip is scanned, can also be measured

Figure 1.4.17 Large-scale, atomic resolution, topographical STM image of the Si(100)2 × 1 surface showing two kinds of steps and various point defects. This is technologically most important as integrated circuits are made on this silicon surface and understanding the local structure on the atomic scale is important for processing and manufacturing of semiconductor microprocessors. Adapted from [25].

23 24

E. R. Müller (1911–1977), German–American physicist. G. Binning and H. Rohrer shared the 1986 Nobel prize in Physics with E. Ruska, and they were cited for “their design of the scanning tunneling microscope.”

OUP CORRECTED – FINAL

48 Introduction to Materials Characterization and Analysis (§11.4). The forces between the tip and surface can have various origins, such as electrostatic, van der Waals, magnetic, etc., giving rise to various variants of these AFMs. These scanning probe instruments and their applications in materials analysis (Weisendanger, 1994; Meyer, Hug, and Bennewitz, 2003) are discussed in §11.

1.4.4

Digital Imaging

As electronic acquisition and analysis of images is now routinely used in materials characterization, it is important to have at least a basic knowledge of digital image acquisition, storage, processing, and analysis methods. While there are excellent textbooks (Gonzales and Woods, 2018; Russ, 1992) and review articles [27–29] that an interested reader can consult for a detailed treatment of the subject, a very brief introduction is included here for completeness. The sequence of procedures followed in digital imaging is acquisition (Fig. 1.4.18a), followed by processing, analysis, and output (Fig. 1.4.18b). Typically, an image acquired from a materials characterization source, such as an optical, electron, or scanning probe microscope, is digitized and then stored. By digitization we mean the generation of a digital image, a[m,n], in a 2D discrete space that is derived from an analog image, a(x,y), in a 2D continuous space through a sampling process. The analog image is divided into N rows and M columns and the intersection of a specific row and column, with integer coordinates [n,m], is called a picture element or pixel. The value, a[m,n], assigned to each pixel in the digital image is the average brightness in the pixel, rounded to the nearest integer value, with L different gray levels in a process called amplitude (a)

(b) Image Formation

Illumination Source

Digitization Pre-Processing Image Output Segmentation Output (digitized) Image

Imaging System

Specimen

Internal image plane

Post-Processing Feature Extraction

Data Output

Figure 1.4.18 (a) Schematic illustration of the digital image acquisition process, which include illumination, the specimen, the imaging system and the digitized image. (b) Steps in digital imaging, shown as a flowchart. (a) Adapted from Gonzalez and Woods (2002). (b) Adapted from [27].

OUP CORRECTED – FINAL

Methods of Characterization: Spectroscopy, Diffraction, and Imaging 49 quantization. Usually, L = 2B , where B is the number of bits used to represent the brightness levels in the image; if B = 1, we get a binary (black and white) image, and if B > 1, we get a gray-scale image. While the acquired image can itself be the output, it often suffers from defects such as the inclusion of electronic noise, uneven illumination over the field of view, and/or specimen drift. These defects are corrected in the pre-processing step. Further analysis of the image for quantitative interpretation requires a sophisticated and complex segmentation step, where features of interest in the image are identified and discriminated from the background. Often, the segmentation result requires some post-processing correction and then the image, containing the distribution of the desired object(s), is analyzed in the feature extraction step to obtain a number of quantitative parameters, including size, position, texture, and (when possible and necessary) appropriate statistical analysis. Here, we introduce only the first three important steps of general interest in digital imaging; the final two steps of post-processing and feature extraction are specific to the particular imaging methods and the characterization problem at hand, and as such are best left for more detailed discussions, readily available in specific textbooks cited earlier. 1.4.4.1

Image Acquisition, Digitization, and Storage

A typical SEM image of a cleaved silicon surface (Fig. 1.4.19) can be used to illustrate the effect of sampling in creating the digital image. The rate at which the signal intensity changes in space is called the spatial frequency; typical features in the image such as edges and small particles represent high spatial frequencies and are best sampled at shorter intervals to improve the quality of the digital image. The sampling frequency in the spatial axis is also known as the resolution; in the intensity axis it is known as quantization. Figure 1.4.20 shows the effects of resolution and quantization on the quality of the digitized image. While it is obvious that the best-quality image is obtained for largest values of resolution and quantization, this choice often leads to very large data files, which reduces its attractiveness.

(a)

(b) 250

Intensity

200

20 μm

150 100 50 0 0

100 200 300 400 500 600 Distance (pixels)

Figure 1.4.19 (a) A scanning electron micrograph of a cleaved silicon wafer, and (b) the intensity trace along the white line shown in the image (a). The image is typically sampled at a specific number of points (black dots) and the digital approximation is shown by the black line. Adapted from [27].

OUP CORRECTED – FINAL

50 Introduction to Materials Characterization and Analysis 20 μm

(a)

(b)

(c)

(d)

Figure 1.4.20 The effect of resolution and quantization on the digital image. The image from Figure 1.4.19, now shown with (a) 64 × 64 pixels and four gray levels, (b) 64 × 64 pixels and 256 gray levels, (c) 512 × 512 pixels and four gray levels, and (d) 512 × 512 pixels and 256 gray levels. Adapted from [27].

The image file size, IFS, is given by IFS = Nx Ny Bpp

(1.4.3)

where Nx and Ny are the number of pixels in the two orthogonal axes, and Bpp , which depends on the quantization, is the number of bytes occupied by each pixel. By definition, a byte is composed of 8 binary digits (bits) and a full byte represents 256 values. For examples, a typical gray-scale image with 256 levels corresponds to a digital value ranging from 0 (black) to 255 (white), will require 1 byte/pixel. Such digital images or arrays of numbers can be stored in different file formats, with different compression protocols developed to reduce the file size. TIFF (Tagged Image File Format) is the most flexible format that can be easily exchanged between different computer platforms, and which accepts different kinds of lossless compression methods. It is also the format of choice for many characterization applications. Alternatively, JPEG (Joint Photograph Expert Group) is another standard and commonly used file format that presents images of good visual quality, but often with lossy compression (Fig. 1.4.21). It allows for several levels of data compression, and the higher the level, the smaller the file (more efficient storage) and the greater the loss of information. For scientific and technical purposes, it is important to retain the precision and details of the acquired image and therefore it is best to avoid the JPEG format since the compressions compromise the data. 1.4.4.2

Pre-Processing: Look-Up Tables, Histogram Equalization, Point, and Kernel Operations

An image displayed on a computer monitor need not be a direct mapping of the original image. Generally, a look-up table (LUT) is used to map the image

OUP CORRECTED – FINAL

Methods of Characterization: Spectroscopy, Diffraction, and Imaging 51

(a)

(b)

(c)

Figure 1.4.21 Storing SEM images of micrometer-size tin spheres. (a) A .TIF file requiring 306 kB of storage. The same file enlarged by a factor of four (4×) using the original image stored as (b) .TIF, and (c) .JPEG requiring only 42 kB of storage. Note that the .JPEG image is degraded and pixelated compared to the .TIF image. Adapted from Goldstein et al. (2003).

intensity values to the brightness values in the display (Fig. 1.4.22a). If the LUT is linear with unit slope and zero intercept the image is directly mapped to the display. The most common example of a LUT operation is the one used to control the brightness (Fig. 1.4.22b) and contrast (Fig. 1.4.22c) in the displayed image. The functional form of the linear LUT is g(u) = u+b with image values u ∈ [0, 1]. When b > 0 (< 0), we get a brightening (darkening) of the image. In addition, the contrast can also be manipulated using a LUT of the form g(u) = mu + b, where b = (1 − m) /2. Then the contrast is enhanced with a large slope, m (→ ∞), and negative intercept, b (→ −∞); alternatively, the contrast is reduced with a small slope, m (→ 0) and positive intercept, b → 1/2. Note that the image is inverted when m = −1, b = 1. Finally, a nonlinear LUT of the form g(u) = uγ , also known as the gamma correction, is often used (Fig. 1.4.22d). Now, when γ < 1, the function approximates a logarithmic LUT with contrast expansion in the dark region, and when γ > 1, it approximates an exponential LUT with contrast reduction in the dark region. In practice, the intensity values of a typical image, plotted as a histogram, are often clustered around the mid-gray value and fall off on either side (Fig. 1.4.21e, top). Such a clustered histogram indicates that the contrast of the image is not maximized. This image is normally transformed so that the distribution of intensity values is such that all intensity values are equally represented in the image. This is known as histogram equalization (Fig. 1.4.21e, bottom), and helps optimize image contrast. Modern image acquisition software includes a live window that includes the image and its histogram, and it can be optimally adjusted by changing the illumination and/or exposure time. Needless to say, it can also be adjusted off-line, post image acquisition. In general, the LUT operations discussed so far are classified as point operations, where the intensity, IO [m,n], of the output image at any pixel with coordinates

OUP CORRECTED – FINAL

52 Introduction to Materials Characterization and Analysis

Bright

High 15000 10000 5000

γ=2

Dark Bright

γ = 0.5

Low

white High

γ >1 Dark 255

(a) Look-up table

(b) Brightness

(c) Contrast

100

200

0

100

200

15000

5000

Low

0

0

10000

γ 1) or decrease (γ < 1) contrast. (e) A histogram of the number of pixels in the image at every pixel intensity (0 to 255) is typically clustered (top) around mid-value. By equalizing or stretching the histogram over the entire intensity range, typically improves the contrast in the image. Adapted from Farid (2010).

[m,n] is dependent only on the input image, II [m,n], at the same coordinates. i.e. the two are related by the function, F, such that IO [m, n] = F (II [m, n])

(1.4.4)

Such point operations readily lend themselves to algebraic and logic operations on two or more images, in such a way that

OUP CORRECTED – FINAL

Methods of Characterization: Spectroscopy, Diffraction, and Imaging 53 IO [m, n] = F (I1 [m, n] , I2 [m, n])

(1.4.5)

where the function, F, relates the intensities at the same pixel position [m,n] in the two input images. The simplest algebraic functions are addition, subtraction, multiplication, and division. Addition is often used to improve the signal-to-noise ratio (SNR): several fields of the same image are acquired and added together to give an improvement of SNR by the square root of the number of fields. Alternatively, subtraction is used to eliminate contributions from noise or uneven background as may be expected for poor or uneven illumination in an optical microscope (Fig. 1.4.23). Multiplication is used to change pixel intensities using another image as a mask, and division is sometimes used to normalize the response of a CCD camera. Neighborhood operations differ from point operations, in that their output intensity at any pixel position depends not only on the input intensity at the same pixel position but also on the intensities of its neighbors. Also known as kernel operations, they are critically important for applications, which include noise filtering, background subtraction, and edge detection. The example, Figure 1.4.24(i), uses a neighborhood—which is always odd-sided and whose weights are represented as a matrix of 3 × 3 pixels. The intensity of each pixel is multiplied by a certain weight (kernel), the results are summed together and divided by the total weight for all pixels, and the result is then plotted as the output for the neighborhood center. Kernels are of two types: (a) those with only positive values, called low-pass filters, because they reduce the high-frequency component of the images, and result in reducing noise and blur in the image, and (b) those with mixed positive and negative values, called high-pass filters, because they increase high spatial frequencies in the image, and result in sharpening the image but at the same time increasing the noise. These two filters are also illustrated in Figure 1.4.24(ii), and their effects on an image, i.e. the blurring effect of a low-pass filter, and

(a)

(b)

(c)

Figure 1.4.23 Background subtraction. (a) An optical micrograph of hypereutectic cast iron at 200× BF, 1300 ×1030 pixels that is unevenly illuminated. (b) The estimated background image intensity. (c) The background subtracted image. Adapted from [27].

OUP CORRECTED – FINAL

54 Introduction to Materials Characterization and Analysis (i)

(ii) (a) 5 5 5 5 5 5

(b)

Input

5 5 5 5 5 5

5 5 5 5 5 5

5 5 5 5 5 5

5 5 5 5 5 5

5 5 5 5 5 5

9 9 9 9 9 9

9 9 9 9 9 9

9 9 9 9 9 9

9 9 9 9 9 9

9 9 9 9 9 9

1 1 1 1 1 1 1 1 1

9 9 9 9 9 9

5 5 5 5

–1 –1 –1 –1 8 –1 –1 –1 –1

0 0 0 0

6 6 6 6

8 8 8 8

9 9 9 9

–12 12 –12 12 –12 12 –12 12

0 0 0 0

Output

Figure 1.4.24 (i) Sequence used in a neighborhood operation, with the neighborhood analyzed on the left and the calculated output pixel on the right. The process starts at the top and is carried out, column by column, across a line and then line by line, until the entire image is processed. (ii) Application of low-pass and high-pass filter to a simple image. (a) Original image and line profile (left), the low-pass kernel (center), and the output image and line profile (right). (b) Original image and line profile (left), the high-pass kernel (center) and the output image and line profile (right). Adapted from [27].

the sharpening (edge enhancement) of the high-pass filter, are illustrated in Figure 1.4.25. Now, consider the filters (kernels) shown in Figure 1.4.26a, which are directional filters that have different effects in different image directions. As it turns out, these kernels are low-pass filters in one direction, and high-pass filters in the orthogonal direction. Hence, they enhance edge directionality, and approximate the derivative of the image in x- and y-directions. In fact, they are the basis of the Sobel edge detector, which is often used in segmentation, and corresponds to the intensity gradient in the image. It is given by  Sobel [I (x, y)] =



∂I (x, y) ∂x

2

 +

∂I (x, y) ∂y

2 (1.4.6)

OUP CORRECTED – FINAL

Methods of Characterization: Spectroscopy, Diffraction, and Imaging 55 (a)

(b)

(c)

40 μm Figure 1.4.25 The effect of low-pass and high-pass filters. (a) The original image. The effect of (b) a 9 × 9 kernel low-pass filter, and (c) a 3 × 3 kernel high-pass filter. Adapted from [27].

(a)

(c)

1

0

–1

1

2

1

2

0

–2

0

0

0

1

0

–1

–1

–1

–1

∂

∂

∂x

∂y (d)

(b) 50 μm

(e) Figure 1.4.26 (a) The kernels for partial x- and y-derivatives. (b) The original image. (c) x-partial derivative image. (d) y-partial derivative image, and (e) Magnitude of the image after the Sobel operation. Adapted from [27].

where the partial derivatives correspond to the two kernels in Figure 1.4.26a. The effect of these operations on an image is shown in Figure 1.4.26b–e; in particular, the Sobel operator enhances the edges while the uniform regions become black, and can be used in the segmentation step to discriminate objects in an image. 1.4.4.3

Image Segmentation

This is the most complex step in the digital imaging flow-chart illustrated in Figure 1.4.18b, because it requires the computer to perform cognitive functions similar to the human brain. The latter performs a rapid processing of numerous inputs—specific shape, texture, brightness, boundaries—in association with previous experience, but unfortunately, computers do not have such associative

OUP CORRECTED – FINAL

56 Introduction to Materials Characterization and Analysis

Figure 1.4.27 (a) An image of catalyst particles. (b) The histogram of intensities and a highlighted threshold level. (c) The segmentation obtained from the thresholding operation. Adapted from [27].

(b) Number of pixels

(a)

(c) threshold

400

200

0

0

50

100 150 200 250 Pixel intensity

power, and the classification of an object is done by classifying each pixel of the image as belonging to the object, or not. The simplest way to segment is to work with the intensities of the pixels; it is considered to be part of the object if its brightness is above a certain level. In the simplest process of intensity thresholding, the choice of the threshold, T, is based on the image histogram, i.e. if I (x,y) > T, then the pixel at (x,y) belongs to the object class (Fig. 1.4.27). Alternatively, segmentation can also be based on the contours, and the simplest method is to apply the Sobel edge detector. If such edges form closed boundaries, they can be clearly identified with the object, but if the boundaries or contours are incomplete, it limits the ability to detect the objects of interest. Several other contour-based methods, such as Marr-Hildreth [30] and Canny [31] methods, have been developed to overcome these limitations; however, further details are beyond the introductory nature of this discussion. The practice of image processing cannot be defined in general terms, as so much of it relates to the actual tasks in mind. For instance, one can include in image processing the task of modeling the background and then subtracting it, which is routinely done in EELS spectroscopy (§9.4.2.3). Examples include the coloring of an image using a well-defined scheme (Fig. 6.8.12), or according to the atomic species in quantitative chemical mapping using EDXS (Fig. 1.4.15). On the other hand, techniques such as particle measurements, closing, erosion, area, moment, counting, and so on are incredibly important at lower resolution, but rarely (if ever) used in high-resolution TEM (§9.3.5). Then, there are the techniques of filters and convolutions, both in real and reciprocal space, which have been introduced, but again, in practice, must be tailored to the actual “question” in mind. Finally, image processing has been covered in all its detail in specialized textbooks (Gonzales and Woods, 2002), but here our interest is limited to providing basic information of general use in characterization.

1.4.5 In Situ Methods Across Spatial and Temporal Scales The emphasis of this book is to describe principles and applications of characterization methods that are broadly applied to studying static and motionless

OUP CORRECTED – FINAL

Features of Materials Used for Characterization 57 states of materials. These studies include structural and analytical methods that have provided incredible details of the structure, form, and function of matter, nature, and life. However, matter reacts, undergoes transformations, and routinely exhibits changes in its thermodynamic state. These changes in equilibrium are often unstable and transient in nature. As a result, there is substantial interest in “seeing” how the building blocks of nature, if possible at the atomic length scale, react to external forces, e.g. changes in the thermodynamic state of the material. In other words, a fourth dimension, time, now becomes important in characterizing the dynamic behavior of materials. Even though time-resolved studies are not new, and date back to the study of galloping horse in the nineteenth century [36], recent developments permit in situ studies at the atomic-length scale and even at femtosecond time scales. In practice, there are many hurdles to carry out these dynamic studies even at slower time scales, much of which revolves around the nature of the specimen, their preparation, and the appropriate specimen environment. We provide very brief introductions to these dynamic methods of materials characterization in this book for X-ray (§7.11.5), neutron (§8.8.3), and electron (§9.5.5) probes, but for those interested in further details, the monograph edited by Ziegler et al. (2014) gives a good overview of time-resolved and in situ methods, providing a snapshot of these rapidly developing techniques with particular emphasis on parameters that limit their application.

1.5 Features of Materials Used for Characterization In general, two principal features of materials, i.e. their electronic structure, including their atomic mass, and crystallography, are used for characterization and are highlighted throughout this text. For spectroscopy, we probe different aspects of the electronic structure of the material; this can be done starting with the energy levels of its constituent atoms (§2). The core levels, or the inner-shell electron energy levels, are barely perturbed by atomic bonding, and continue to provide valuable chemical information, even in bulk form, for most materials. Techniques that are based on probing such energy levels and related transitions, including the design of simple detectors for measuring X-rays and electrons, are described first (§2). The electronic structure of materials, especially the outer levels, change when the atoms bind into molecules and solids. In addition, molecules exhibit vibrational and rotational modes, whereas crystalline solids develop a bulk band structure, conveniently described by an itinerant or delocalized electron model (Sutton, 1993). The solid also develops acoustic or phonon modes of excitation. Following a conceptual introduction to the relevant electronic structure of molecules and solids, various techniques that are based on probing their rich electronic structures are introduced in §3.

OUP CORRECTED – FINAL

58 Introduction to Materials Characterization and Analysis Whenever electromagnetic radiation interacts with matter, some form of scattering always takes place. In particular, the type of scattering that we consider can be simply thought of as arising from the absorption of the incident radiation followed by its re-emission. This model lends itself to both elastic (Rayleigh) and inelastic (Stokes and anti-Stokes) scattering for molecules and atomic arrangements in solids (§3). An electromagnetic wave, such as an X-ray, can be described by a propagating electric field, which varies sinusoidally with time (Fig. 1.3.4). At any given location, when such an oscillating electric field encounters a charged particle, e.g. an electron in an atom, it will set it also in oscillatory motion about its mean position. An oscillating or accelerating electric charge will emit an electromagnetic radiation. Thus, the atomic electron will not only absorb the incoming radiation but will also re-emit it with the same frequency ( f ) and wavelength (λ). This process is called scattering, and in X-ray diffraction it is characterized by the reemitted radiation being coherent with the incident wave. By coherent, we mean a constant phase difference, φ, between the incident and scattered radiation, such that in the case of X-rays, φ = λ/2. Even if the incident radiation is a parallel beam, the scattered radiation is re-emitted in all directions, with intensities dependent on the angle of scattering (we discuss such Thomson scattering in §7.2.1). Note that in this simple model, the contributions from the positively charged nucleus can be ignored because its mass is much larger (∼1,000 times) than that of the electron, and as such, it cannot be oscillated to any significant degree by the incident wave. The collection of all electrons in an atom will thus scatter the incident wave in all directions (Fig. 1.4.10). However, for a crystalline material, where atoms are arranged in a well-defined periodic array, their collective scattered radiations interfere constructively only along certain well-defined direction determined by Bragg law, (1.4.1), but destructively interfere for all other directions (§7.4). Further, if the crystal is not perfect, i.e. it includes defects, then the requirement for complete destructive interference is not met, and intensities can be found at angles other than the Bragg angle. From this brief discussion, it is important to recognize that before we present the physics of diffraction, we need to develop a technical vocabulary to describe the symmetry and periodic arrangement of atoms in crystals as well as their defects. Thus, in §4, we begin with an introduction to crystallography and the general principles of diffraction. We then discuss details of the different kinds of probes and scattering of ions and ion-based characterization methods (§5), and present an introduction to optics, microscopy, and ellipsometry (§6). In the second half of the book, we discuss diffraction in detail for X-rays (§7), and electrons and neutrons (§8). Finally, we discuss imaging with electrons in transmission (§9) and scanning (§10) modes, as well as scanning probe methods (§11), and conclude with a comprehensive set of tables (§12) comparing the different spectroscopy, imaging, and diffraction methods discussed throughout the book.

OUP CORRECTED – FINAL

Further Reading 59

Summary Optimizing materials microstructures is central to materials development, enabling new technologies, and in a broad sense, impacting every aspect of our life today. Central to this exercise is materials characterization, which helps relate synthesis and processing of materials to their structure, properties, and performance. The reach of materials characterization is indeed broad and includes, for example, optimization of materials microstructures for advanced technologies, unraveling the structure of biological molecules and complexes, failure analysis, understanding the morphology of rocks and geological processes, and even in the analysis, conservation, and authentication of sculptures or paintings. Materials characterization typically uses probe and signal radiations to interrogate a specimen. The probe and signal radiations may or may not be the same, and the interactions between the probe and the specimen may be elastic or inelastic, coherent or incoherent. The characterization techniques can be broadly classified as spectroscopy (involving absorption, emission, and/or transitions processes), diffraction and scattering, and imaging and microscopy. The probes are broadly based on the electromagnetic spectrum, and their characteristics (energy, wavelength, momentum, polarization) define their interaction with matter, and determine the nature, scope, and details of the possible methods of characterization. In addition, the probes or signals can also be ions or neutrons. Principal features of the materials, i.e. details of their electronic structure, including atomic mass, composition, and crystallography, contribute to the observable signals and define possible characterization methods that are explored systematically in later chapters. This introductory chapter provides some basic background, including definitions, on the underlying physics of spectroscopy, diffraction and imaging, interactions of radiations with matter, and a rudimentary introduction to digital imaging. Further, it provides numerous motivational examples of characterization in a variety of different contexts. .............................................................................. FURTHER READING

Brandon, D., and W. D. Kaplan. Microstructural Characterization of Materials. Chichester: Wiley, 2008. Brooks, C., and A. Choudhury. Failure Analysis of Engineering Materials. New York: McGraw-Hill, 2001. Brundle, C. R., C. A. Evans and S. Wilson (eds.). Encyclopedia of Materials Characterization. Stoneham: Butterworth-Heinemann, 1992. Callister, W. D., Jr., and D. G. Rethwisch. Materials Science and Engineering: An Introduction. New York: Wiley, 2009.

OUP CORRECTED – FINAL

60 Introduction to Materials Characterization, Analysis, and Metrology Farid, H. Fundamentals of Image Processing. Hanover: Dartmouth College Press, 2010. Feldman, L. C., and J. W. Mayer. Fundamentals of Surface and Thin Film Analysis. New York: North-Holland, 1986. Flewitt, P. E. J., and R. K. Wild. Physical Methods of Materials Characterization. Bristol: IoP Press, 2003. Goldstein, J., D. Newbury, D. Joy, C. Lyman, P. Echlin, E. Lifshin, L. Sawyer, and J. Michael, Scanning Electron Microscopy and X-Ray Microanalysis. New York: Springer, 2003. Gonzalez, R. C., and R. E. Woods. Digital Image Processing. Upper Saddle River: Prentice-Hall, 2002. Hecht, E. Optics. Reading: Addison Wesley, 2002. Hüfner, S. Photoelectron Spectroscopy. Berlin: Springer-Verlag, 2003. Krishnan, K. M. Fundamentals and Applications of Magnetic Materials. Oxford: Oxford University Press, 2016, Chapter 8. Kurzydlowski, K. J., and B. Ralph. The Quantitative Description of the Microstructure of Materials. Boca Raton: CRC Press, 1995. Meyer, E., J. J. Hug, and R. Bennewitz. Scanning Probe Microscopy—The Lab on a Tip. Berlin: Springer-Verlag, 2003. Miller, M. K., A. Cerezo, M. G. Hetherington, and G. D. W. Smith. Atom Probe Field Ion Microscopy. Oxford: Oxford University Press, 1996. Owens, F. J., and C. P. Poole Jr., The Physics and Chemistry of Nanosolids. Wiley, 2008. Russ, J. C. The Image Processing Handbook. Boca Raton: CRC Press, 1992. Sheppard, C. J. R., C. Sheppard, and D. Shotton. Confocal Laser Scanning Microscopy. Abigndon: BIOS Scientific Publishers, 1997. Sherwood, P. M. A. Vibrational Spectroscopy of Solids. Cambridge: Cambridge University Press, 1972. Somorjai, G. A. Chemistry in Two Dimensions: Surfaces. New York: Cornell University Press, 1981. Stokes, D. E. Pasteur’s Quadrant: Basic Science and Technological Innovation. Washington, DC: Brookings Institution Press, 1997. Sutton, A. P. Electronic Structure of Materials. Oxford: Oxford University Press, 1993. Vernon, R. H. A Practical Guide to Rock Microstructure. Cambridge: Cambridge University Press, 2004. Vickerman, J. C., ed. Surface Analysis—The Principal Techniques. Chichester: Wiley, 1997. Wiesendanger, R. Scanning Probe Microscopy and Spectroscopy. Cambridge: Cambridge University Press, 1994. Williams, D., A. R. Pelton, and R. Gronsky, eds. Images of Materials. Oxford: Oxford University Press, 1991. Woodruff, D. P, and T. A. Delchar. Modern Techniques of Surface Analysis. Cambridge: Cambridge University Press, 1986. Ziegler, A., H. Graafsma, X. F. Zhang, and J. W. M. Frenken, eds. In-Situ Materials Characterization Across Spatial and Temporal Scales. New York: Springer, 2014.

OUP CORRECTED – FINAL

References 61 .............................................................................. REFERENCES

[1] Materials Science and Engineering for the 1990s: Maintaining Competitiveness in the Age of Materials, Washington, DC: The National Academy Press, 1989. A pdf copy of the report is available from http://nap.edu/758 [2] Long, H., S. Mao, Y. Liu, Z. Zhang, and X. Han. “Microstructural and compositional design of Ni-based single crystalline superalloys — A review.” Journal of Alloys and Compounds 743 (2018): 203–20. [3] Xiang, S., S. Mao, H. Wei, Y. Liu, J. Zhang, Z. Shen, H. Long, H. Zhang, X. Wang, Z. Zhang, and X. Han “Selective Evolution of Secondary γ  Precipitation in a Ni-Based Single Crystal Superalloy Both in the γ Matrix and at the Dislocation Nodes.” Acta Materialia 116 (2016): 343–53. [4] Liu, M., and S. Nemat-Nasser. “The Microstructure and Boundary Phases of In-Situ Reinforced Silicon Nitride.” Materials Science and Engineering A 254, (1998): 242–52. [5] Clarke, D. R. “Grain Boundaries in Polycrystalline Ceramics.” Annual Review of Materials Science 17 (1987): 57. [6] Watson, J. D., and F. H. C. Crick. “Molecular Structure of Nucleic Acids: A Structure for Deoxyribose Nucleic Acid.” Nature 171 (1953): 737–8. [7] Franklin, R. E., and R. G. Gosling. “Evidence for 2-Chain Helix in Crystalline Structure of Sodium Deoxyribonucleate.” Nature 172 (1953): 156–7. [8] Felkins, K., H. P. Leighly, and A. Jankovic. “The Royal Mail Ship Titanic: Did a Metallurgical Failure Cause a Night to Remember?” Journal of Metals 1998: 12–17. [9] Götze, J. “Potential of Cathodoluminescence (CL) Microscopy and Spectroscopy for the Analysis of Minerals and Materials.” Analytical and Bioanalytical Chemistry 374 (2002): 703. [10] Carson, W. D., and C. Denison. “Mechanisms of Porphyroblast Crystallization: Results from High-Resolution Computed X-ray Tomography.” Science 257 (1992): 1236. [11] Whitehead, M. E., A. S. A. Karim, M. H. Loretto and R. E. Smallman. “Electron Radiation Damage in H.C.P. Metals—II. The Nature of the Defect Clusters in Zn and Cd Formed by Irradiation in the HVEM.” Acta Metallurgica 26, no. 6 (1977): 983–93. [12] Li, M., M. A. Kirk, P. M. Baldo, D. Xu and B. D. Wirth. “Study of Defect Evolution by TEM with In Situ Ion Irradiation and Coordinated Modelling.” Philosophical Magazine 92, no. 16 (2012): 2048–78. [13] Krause, M. O. “Atomic Radiative and Radiationless Yields for K and L Shells.” Journal of Physical and Chemical Reference Data 8 (1979): 307. [14] Raman, C.V., and K. S. Krishnan. “The Production of New Radiations by Light Scattering. Part I.” Proc. Roy. Soc. 122, no. 789 (1929): 23–35. [15] Nutall, R. H. “The First Microscopes of Henry Clifton Sorby.” Technology and Society 22 (1981): 275–80. [16] Montoya, S.A., S. Couture, J. J. Chess, J. C. T. Lee, N. Kent, D. Henze, S. K. Sinha, M.-Y. Im, S. D. Kevan, P. Fischer, B. J. McMorran, V. Lomakin,

OUP CORRECTED – FINAL

62 Introduction to Materials Characterization, Analysis, and Metrology

[17]

[18]

[19] [20] [21] [22] [23]

[24]

[25]

[26] [27]

[28]

[29] [30]

[31] [32]

S. Roy, and E. E. Fullerton. “Tailoring Magnetic Energies to Form Dipole Skyrmions and Skyrmion Lattices.” Physical Review B 95 (2017): 024415. Thorel, A., J Ciston, T. P. Bardel, C. -Y. Song, and U. Dahman. “Observation of the Atomic Structure of ß’-SiAlON Using Three Generations of HighResolution Electron Microscopes.” Philosophical Magazine A 93(2013): 1172–81. Müller, E. R., and K. Bahadur. “Field Ionization of Gases at a Metal Surface and the Resolution of the Field Ion Microscope.” Physical Review 102 (1956): 624. Miller, M. K., T. F. Kelly, K. Rajan, and S. P. Ringer. “The Future of Atom Probe Tomography.” Materials Today 15, no. 4 (2012): 158–65. Amouyal, Y., and G. Schmitz. “Atom Probe Tomography—A Cornerstone in Materials Characterization”. MRS Bulletin 41 (2016): 13–18. Muller, E. W., J. A. Panitz, and S. B. McLane. “The Atom-Probe Field Ion Microscope.” Review of Scientific Instruments 39 (1968): 83. Panitz, J.A. “The 10 cm Atom Probe.” Review of Scientific Instruments 44 (1973): 1034. Cerezo, A., T. J. Godfrey and G. D. W. Smith. “Application of a PositionSensitive Detector to Atom Probe Microanalysis.” Review of Scientific Instruments 59 (1988): 862. Kelly, T. F., T. T. Gribb, J. D. Olson, R. L. Martens, J. D. Shepard, S. A. Wiener, T. C. Kunicki, R. M. Ulfig, D. R. Lenz, E. M. Strennen, E. Oltman, J. H. Bunton, and D. R. Strait. “First Data from a Commercial Local Electrode Atom Probe (LEAP).” Microscopy and Microanalysis 13 (2004): 373–83. Hamers, R. J., R. M. Tromp, and J. E. Demuth. “Electronic and Geometric Structure of Si(111)-(7 × 7) and Si(001) Surfaces.” Surface Science 181 (1987): 346–55. Binnig, G., C. F. Quate and Ch. Gerber. “Atomic Force Microscope.” Physical Review Letters 56 (1986): 930. Paciornik, S., and M. H. P. Mauricio. “Digital Imaging.” In: ASM Handbook, Metallography and Microstructures, edited by G. F. Vander Voort, Vol. 9, 368–402. Materials Park: ASM International, 2014. Russ, J. C. “Fundamentals of Image Processing and Measurement.” In Encyclopedia of Computer Science and Technology, 2nd ed., edited by P. A. Laplate. Boca Raton: Taylor & Francis Group, 2016. Russ, J. C. “Image Analysis of Foods.” Journal of Food Science 80, no. 9 (2015): 1974–87. Marr, D., and E. Hildreth. “Theory of Edge Detection.” Proceedings of the Royal Society of London. Series B, Biological Sciences 207, no. 1167 (1980): 187–217. Canny, J. “A Computational Approach to Edge Detection.” IEEE Transactions on Pattern Analysis and Machine Intelligence 8 (1986): 679–698. Feynman, R. P. “There’s Plenty of Room at the Bottom (Data Storage).” Journal of Microelectromechanical Systems 1, no. 1 (1992): 60–66. doi:10.1109/84.128057. This is a transcript of the talk by Richard Feynman.

OUP CORRECTED – FINAL

Exercises 63 [33] Bednorz, J. G., and K. A. Müller. “Possible High T c Superconductivity in the Ba−La−Cu−O System.” Zeitschrift für Physik B Condensed Matter 64 (1986): 189–93. [34] Zhao, J. C., and J. H. Westbrook. “Ultrahigh-Temperature Materials for Jet Engines.” MRS Bulletin 28, no. 9 (2003): 622–30. [35] Seah, M. P., and W. A. Dench. “Quantitative Electron Spectroscopy of Surfaces: A Standard Data Base for Electron Inelastic Mean Free Paths in Solid.” Surface and Interface Analysis 1, no. 1 (1979): 2–11. [36] Eadweard Muybridge collections, Stanford University Libraries. Available at: https://library.stanford.edu/collections/eadweard-muybridgephotographs. .............................................................................. EXERCISES

A. Test Your Knowledge There may be more than one, or no, correct answer. 1. Metrology is (a) the study of measurements. (b) the study of weather patterns. (c) not relevant for good materials characterization. 2. Microstructure (a) includes atomic, mesoscopic, and microscopic length scales. (b) is related to materials properties and function. (c) can be observed by the naked eye. 3. Materials characterization involves (a) probes and signals. (b) probes that may damage some materials. (c) probes and signals that are always different radiations. 4. Resolution in the context of materials characterization is (a) a measure of how determined you are to get the job done. (b) only given by the penetration depth of the probe. (c) divided into spatial, depth, and temporal categories. 5. The interaction of a probe with the specimen is (a) elastic if it does not lose any energy. (b) coherent if probe and signal have the same phase. (c) coherent if all the signal is uniformly of the same phase. 6. The electromagnetic spectrum (a) provides a rich source of probes for materials analysis. (b) includes scanning probes. (c) can be displayed in terms of energy, frequency, or wavelength. (d) matches with critical length scales that describe the microstructure of materials.

OUP CORRECTED – FINAL

64 Introduction to Materials Characterization, Analysis, and Metrology 7. Light (a) is always unpolarized. (b) is a transverse wave with E and H propagating in orthogonal planes. (c) can be linearly, elliptically or circularly polarized. 8. A photon is a quantum of light that (a) behaves as a particle. (b) behaves as a wave packet. (c) can be divided into smaller units of energy. 9. Interaction of a probe with matter (a) will generally not perturb the material. (b) can cause damage. (c) is characterized by a penetration depth and a mean free path length. 10. The Planck constant (a) is used to strengthen your abs. (b) is a very large number. (c) relates the energy of a photon to its frequency. (d) can be used to calculate the wavelength of an object if its velocity is known. 11. Spectroscopy can involve (a) absorption, emission, and transition processes. (b) non-radiative processes. (c) energy levels associated with molecular vibrations and rotations. 12. X-ray diffraction (a) of crystals involves Bragg law. (b) is dominated by the interaction only with atomic electrons. (d) includes contributions from the electrons, protons and neutrons of the atoms. 13. The resolution of a microscope (a) is given by the Rayleigh criterion. (b) can be improved by reducing the wavelength of the probe. (c) can be improved by increasing the refractive index of the medium. 14. Images with atomic resolution can be obtained by ____________ microscopy. (a) optical (b) scanning probe (c) scanning tunneling (d) high resolution transmission electron (e) field ion (f) scanning electron 15. A digitized image (a) generates an image in discrete space. (b) is often derived from an image in continuous space. (c) typically involves a sampling process.

OUP CORRECTED – FINAL

Exercises 65 16. A pixel (a) stands for a picture element. (b) has an averagebrightness value. (c) has binary values (1 or 0) for a grey scale image. 17. In a digital image, the sampling frequency (a) in the spatial axis gives the resolution. (b) in the intensity axis is known as quantization. (c) depends on the number of specimens imaged. 18. File formats in digital imaging include (a) TIFF, which is flexible and involves loss-free compression. (b) JPEG, which is common but has lossy compression. (c) no other formats because they are not discussed here. 19. A look-up table (a) maps the image intensities to the display intensities. (b) if linear, can routinely manipulate contrast and brightness. (c) can be non-linear. (d) operation, in general, is NOT a point operation. 20. Kernel operations in digital imaging (a) are neighborhood operations. (b) only allow low-pass filters. (c) can produce derivatives of images.

B. Problems 1. What do you understand by the term “microstructure of materials?” Write a brief (50–100 word) definition of “microstructure.” 2. What is meant by “materials characterization?” In your own words, briefly (50–100 word) define “materials characterization.” 3. What is meant by “sensitivity” or “detection limit” of a characterization technique? 4. Is it possible to damage a material in the process of characterizing it? Explain, in your own words, using appropriate examples. 5. What is the difference between spatial and temporal resolution? 6. A dust particle of mass 1 μg travels at a velocity of 25,000 kph. Calculate its wavelength. 7. What is the kinetic energy in Joules of an electron accelerated from rest through a potential of 5 kV? 8. What is the de Broglie wavelength of (i) hydrogen atoms (m = 1.67 × 10−27 kg) moving with velocity = 103 m/s? (ii) electrons accelerated by 5,000 V in a SEM? (iii) tennis balls (100 g) traveling with a velocity of 20 m/s? (iv) 4 He+ ions accelerated through 1 MV for RBS experiments?

OUP CORRECTED – FINAL

66 Introduction to Materials Characterization, Analysis, and Metrology 9. If the wavelength of an electron is infinite, the electron must be stationary. True or false? Justify your answer. 10. Please do some general reading and complete as much of the following table as possible. Characterization Technique

Probe in

Probe out

Resolution, sensitivity . . ..

Depth probed

Highlights or comments

Light Microscopy Scanning electron microscopy (SEM) Transmission electron microscopy (TEM) Energy dispersive X-ray spectrometry (EDS) Auger electron spectrometry (AES) X-ray photoelectron spectroscopy (XPS) Fourier transform infrared spectroscopy (FTIR) Raman spectroscopy Ultraviolet and visible spectrometry (UV-Vis) Rutherford back-scattering spectroscopy (RBS) X-ray diffraction (XRD) X-ray fluorescence (XRF) Scanning tunneling microscopy (SPM) Atomic force microscopy (AFM) Nuclear magnetic resonance (NMR) Secondary ion mass spectrometry (SIMS) Inductively coupled plasma mass spectrometry (ICP-MS) Field ion microscopy 11. For CuKα radiation, how many photons are required to deliver 1 Joule of energy? What is the flux (photons/s) required to deliver a power of 1 watt?

OUP CORRECTED – FINAL

Exercises 67 12. X-rays typically used in diffraction have a wavelength of 1.5 Å. What is the energy and momentum of this X-ray photon? What would be the velocity of a corresponding (i) electron, and (ii) proton, that has the same momentum? 13. Read Richard Feynman’s famous article “There is Plenty of Room at the Bottom” [32], considered by many as the first lecture on the promise of nanoscience and nanotechnology. In this talk, he first outlines how all the world’s data can be stored on a speck of dust. He then makes the case for developing the finest electron microscopes to “see” atoms. He then, with characteristic certainty, says: “It would be very easy to make an analysis of any complicated chemical substance; all one would have to do would be to look at it and see where the atoms are.” Now, examine the case of high-T c oxide superconductors [33], for which its discoverers, J. G. Bednorz and K. A. Müller, received the Nobel Prize in 1987, and address the following questions: (i) Is the structure, including the position of atoms in the perovskite unit cell, for these high-T c YBa2 Cu3 O7 oxides well known? (ii) Has this led to an understanding of why these materials are superconducting? (iii) Does this prove or refute the arguments of Professor Feynman? Please explain.

OUP CORRECTED – FINAL

Atomic Structure and Spectra

2 2.1 Introduction

69

2.2 Atomic Structure

69

2.3 Atomic Spectra: Transitions, Emissions, and Secondary Processes

78

2.4 X-Rays as Probes: Generation and Transmission of X-Rays

97

2.5 X-Rays as Signals: CoreLevel Spectroscopy with X-Rays

(a)

(b)

(c)

107

2.6 Surface Analysis: Spectroscopy with Electrons

119

2.7 Select Applications

137

Summary

139

Further Reading

140

References

142

Exercises

143 (d) Iron

(e) Chromium

(f) Cadmium

(g) Lead

(h) Barium

(i) Zinc

Characterization and analysis also play a significant role in art conservation, and are often used for authentication, or help rule out forgery when the provenance of the work is questionable. Sometimes such analysis reveals hidden secrets, as illustrated here in Picasso’s Blue Period painting La Miséreuse accroupie. (a) The painting as it is seen today. (b) Mounting the painting for element-specific, spatially-resolved X-ray fluorescence measurements described in §2.5.2.1. (c) Using such core-level spectroscopy revealed, underneath the Picasso painting, a hidden landscape painting by another artist. Element-specific Xray fluorescence images of the pigments (d) Iron (blue), (e) chromium (yellow), (f) cadmium (red, orange), (g) lead (white), (h) barium, and (i) zinc. Adapted from the New York Times, February 21, 2018.

Principles of Materials Characterization and Metrology. Kannan M. Krishnan, Oxford University Press (2021). © Kannan M. Krishnan. DOI: 10.1093/oso/9780198830252.003.0002

OUP CORRECTED – FINAL

Atomic Structure 69

2.1 Introduction Section 1.4.1 presented a brief discussion of absorption, emission, and radiation processes, based on a simple model of the atom to motivate various spectroscopy methods used in materials characterization. This chapter develops the Bohr1 model further and then presents, in a concise form, a quantum- or wavemechanical model of atomic structure. This model overcomes the major flaw of the Bohr model, i.e. electrons in perpetual acceleration emitting no radiation, a behavior contradicting classical radiation physics. Then, using the latter as a working model of the atomic structure, we revisit in further detail the set of radiative and non-radiative processes introduced in §1 for both electron and X-ray incidence. We discuss the emission, transmission, and detection of characteristic X-rays, and then develop a formulism for their use in quantitative microanalysis. We look at two non-radiative processes: Auger electron spectroscopy (AES) with electron incidence, and X-ray photoelectron spectroscopy (XPS) with X-ray incidence, developing a conceptual understanding of these methods. We also discuss the instrumentation required for the detection of X-rays and electrons. Section §2.2 deals with elementary ideas of atomic structure. Those familiar with these concepts, including spin–orbit coupling, may skip it and move directly to §2.3.

2.2 Atomic Structure 2.2.1

Bohr–Rutherford–Sommerfeld Model

Our early understanding2 of the atomic structure, especially the nuclei, come from scattering experiments; the classical experiment carried out by Rutherford,3 using incident α particles, showed the existence of a positively-charged nucleus, surrounded by electrons in circular orbits.4 At that time, it was also known, from extensive spectroscopic measurements, that atoms emit characteristic but narrow lines of radiation with well-defined wavelength (or frequency), and most importantly, in prescribed numerical sequences. Further, the principle of waveparticle duality (§1.3.3), where the electron wavelength, λ, and momentum, p, are related through the de Broglie5 relation, λ = h/p, was well accepted. Incorporating these conceptual ideas, Bohr proposed the early quantum mechanical model of the atom by first equating the Coulombic attractive force, 1 Ze2 −19 C; rest mass, 4πε0 r 2 , experienced by the electron (charge, e = −1.602 × 10 −31 me = 9.11 × 10 kg) in the atom due to the nuclear charge, +Ze, with the centripetal force, me v2 /r. He also introduced the concept of stationary orbits (2πr = nλ) and included the principle of wave-particle duality, to show that the circular electron orbits are not continuous but quantized in angular momentum, me vr = nh/2π, (n = 1, 2, 3, . . . . ), total energy

1 Niels H. D. Bohr (1885–1962) was a Danish physicist who received the 1922 Nobel prize in Physics for “his services in the investigation of the structure of atoms and of the radiation emanating from them.” 2 Arnold J. W. Sommerfeld (1868– 1951) was a German physicist. 3 Ernest Rutherford received the 1908 Nobel prize in Chemistry for “his investigations into the disintegration of the elements, and the chemistry of radioactive substances.” 4 It is now well established that the scattering of high-energy α particles from different nuclei is distinctly different, and the measurement of their intensity and energy distribution, after being scattered by a material, also provides a direct method to determine the elemental composition of the specimen (see §5.4 for a discussion of the technique of Rutherford backscattering spectroscopy, or RBS). 5 Louis de Broglie (1892–1987) was a French physicist who received the 1929 Nobel prize in Physics for “his discovery of the wave nature of electrons.”

OUP CORRECTED – FINAL

70

Atomic Structure and Spectra En = −

me Z 2 e4 1 Z2 = −13.606 2 (eV) 2 2 2 n 8ε0 h n

(2.2.1)

and radius rn =

ε0 h2 πme Ze

2

n2 =

a0 n2 (Å) Z

(2.2.2)

where, me e4 /8ε02 h2 = 13.606 eV, is known as the Rydberg constant, ε0 = 8.85 × 10−12 F m−1 is the permittivity of free space, and h = 4.135 × 10−15 eV s, is Planck’s constant. Note that the radius is commonly expressed in terms of the first Bohr radius, r1 = a0 = 0.529 Å, of the hydrogen atom (Z = 1). The significant problem with the Bohr model is that it contradicts classical radiation physics as the electron in orbit, in spite of its continuous acceleration, is assumed to not emit any radiation. However, the Bohr model does allow for the emission of radiation when the electron transitions from one energy level (ni ) to another (nf ), derived from (2.2.1) as: hf = Ei − Ef =

6 Werner Heisenberg received the 1932 Nobel prize in Physics for “the creation of quantum mechanics, the application of which has, inter alia, led to the discovery of the allotropic forms of hydrogen.” 7

Erwin Schrödinger and P. A. M. Dirac shared the 1933 Nobel prize in Physics for “the discovery of new productive forms of atomic theory.”

me Z 2 e4 8ε02 h2



1 1 − 2 n2f ni



 = −13.6Z 2

1 1 − 2 n2f ni

 (eV)

(2.2.3)

Now, as both ni and nf are integers, numerical sequences are possible for different combinations of ni and nf , and various observed series, such as Balmer (ni = 3, 4, 5, . . . , and nf = 2), Lyman (ni = 2, 3, 4, . . . , and nf = 1), Paschen (ni = 4, 5, 6 . . . , and nf = 3) etc., can be explained by (2.2.3). Sommerfeld improved upon the work of Bohr by introducing non-circular elliptical orbits with the use of a second azimuthal quantum number to define the degree of ellipticity, and explained additional fine structure observed in the spectra. However, not all the predicted emission lines were observed, suggesting that of all the possible quantum states, only some were preselected and determined the allowed transitions. The currently accepted wave-mechanical version of the quantum theory, developed in a decade of unusual creativity with contributions from many scientists, including Heisenberg,6 Schrodinger,7 and Dirac,7 provided a way to go beyond the limitations of the Bohr model. It accurately predicts and matches observed atomic behavior, including observed transitions and emissions, of particular interest to our discussion. Details are found in any text on quantum mechanics (e.g. Rae, 1992; Sutton, 1993), but here we provide a brief summary highlighting the essential physics involved in electron transitions in atoms and related emissions.

2.2.2 Quantum Mechanical Model In the quantum- or wave-mechanical model, the electrons are described in terms of a probabilistic wave-function, (r, t), such that |(r, t)|2 dr gives the probability

OUP CORRECTED – FINAL

Atomic Structure 71

Probability

1.0

0 Distance from nucleus

Orbital Electron

(a)

Figure 2.2.1 A pictorial representation of the (a) Bohr and (b) wave-mechanical model of the spatial distribution of electrons in the atom.

Nucleus

Adapted from Callister and Rethwisch (2010).

(b)

of finding the electron, at any time, t, in a volume dr = dx.dy.dz, located at the position r = (x, y, z). In this model, any electron in the atom no longer occupies a discrete orbital, as in the Bohr model, but rather its position is described by the probability function (Fig. 2.2.1). In classical mechanics, the path or motion of a particle of mass, me , is prescribed by the Newton law of motion; in quantum mechanics the corresponding equation of motion of the electron is the Schrodinger equation, which describes the time evolution of the wave-function, (r, t), under a potential, V(r, t), as −

∂(r, t) 2 2 ∇ (r, t) + V(r, t) (r, t) = i 2me ∂t

(2.2.4)

h , where  is the reduced Planck constant, where, to avoid writing 2π often,  = 2π me is the mass of the electron, and ∇ is the vector gradient. This time-dependent Schrodinger equation, (2.2.4), can be solved by the method of separation of  

variables, (r, t) = ψE (r)T(t), where T(t) can be shown to be T(t) = e Thus (r, t) = ψE (r)exp(−iEt/)

−iEt 

.

(2.2.5)

and the spatially-dependent term, ψE (r), is an eigenfunction of the timeindependent Schrodinger equation −

2 2 ∇ ψE (r) + V (r)ψE (r) = EψE (r) 2me

(2.2.6)

OUP CORRECTED – FINAL

72

Atomic Structure and Spectra 2

Ze , For electrons in a spherically symmetric Coulombic potential, V (r) = − 4πε 0r experienced due to nuclear charge, +Ze, in the atom, the solution to the Schrodinger equation, (2.2.6), also reflect the spherical symmetry, and takes the form m

ψE (r) = ψ(r, θ , φ) = Rnl (r)Yl l (θ, φ) m

8 A parameter-dependent equation with non-vanishing solutions only for particular values (eigenvalues) of the parameter is an eigenvalue equation, and the associated solutions are the eigenfunctions. 9 Wolfgang Pauli received the 1945 Nobel prize in Physics for “the discovery of the Exclusion Principle, also called the Pauli Principle.”

(2.2.7)

where r, θ , φ are spherical coordinates, Yl l (θ, φ) are spherical harmonics, and θ is measured from the z-axis. The first term, Rnl (r), depends only on the distance m of the electron from the nucleus, and the second term, Yl l(θ, φ), takes care of the spherical symmetry. Further, the requirement that this solution be finite, continuous, single valued, and normalized, introduces three quantum numbers, n, l, and ml , i.e. one for each coordinate axis, which characterize the eigenfunctions8 of the electron. Here, n is the principal quantum number, associated with the radial coordinate, Rnl (r), and has allowed integer values, n = 1, 2, 3 . . . . Note that this quantum number, and only this one, is associated with the earlier Bohr model. l is the orbital quantum number that determines the magnitude of the electron √ angular momentum, |L| = l (l + 1), with integer values, l = 0, 1, 2, 3, . . . , n−1, denoting the sub-shell. The letters s(l = 0) , p (l = 1) , d (l = 2) , f (l = 3) , . . . , are also commonly used. ml is the magnetic quantum number, with ml = 0, ±1, ±2, . . . , ±l, that determines the z-component of the angular momentum Lz = ml . Finally, a relativistic treatment completes the quantum mechanical description of the electrons, by introducing a fourth quantum number, s = 12 , such that the √ √ magnitude of the spin angular momentum is |S| = s (s + 1) = 23  and its components along the z-axis are given by ms , where ms = ±1/2. This quantum number, ms , describes the component of the spin angular momentum, which in simple terms, must be oriented with z-component either up (1/2) or down (–1/2). The occupation of electrons in different orbitals defined by these set of four quantum numbers, n, l, ml , and ms , must also satisfy the Pauli9 exclusion principle, which states that no two electrons can have the same set of four identical quantum numbers. The constraints on the four quantum numbers enumerated above, combined with the exclusion principle, imposes a limit on the number of electrons in each orbital. Thus, the first shell (n = 1) can hold two electrons, the second shell (n = 2) can hold two electrons in the l = 0 (s) sub-shell and six electrons in the l = 1 (p) subshell, for a total of eight electrons. The third shell can hold 18 electrons with two in the 3s (n = 3, l = 0, ml = 0, ms = ±1/2), six in the 3p (n = 3, l = 1, ml = ±1, 0, ms = ±1/2), and 10 in the 3d (n = 3, l = 2, ml = ±2, ±1, 0, ms = ±1/2) sub-shells. For the atom with Z = 36 (Kr), where all the shells up to the third are completely filled, its electronic structure can be written as 1s2 , 2s2 , 2p6 , 3s2 , 3p6 , 3d 10 , 4s2 , 4p6 . Note that the available electrons fill states

OUP CORRECTED – FINAL

Energy →

Atomic Structure 73

d

f

d

f

d

p s

f

d

p s

d

p s

p s

p s

Figure 2.2.2 The relative energies of the electrons in the atom occupying various shells and sub-shells. Further, note that for l >1, the spin–orbit coupling leads to further splitting of the sub-shells as discussed in the text, and in Figure 2.2.6.

p s s

1

2 3 4 5 6 Principal quantum number, n →

7

of the lowest energy; such a configuration is also called the ground or unexcited state of the atom. The energies of the electrons are determined by the principal quantum number, n, and to a lesser extent by the orbital quantum number l (Fig. 2.2.2). The periodic table of the elements is built by assigning electrons, as described above, and Table 2.2.1 summarizes the ground state electronic configuration of the first 36 common elements. The orbitals that we have discussed thus far also have an angular character, which is central to directional bonding in molecules and covalent crystals (to be discussed further in §3). This angular character is determined by the angular dependence of the wave functions or the appropriate spherical harmonic in 2 m (2.2.7), with its shape determined by the probability distribution, Yl l (θ, φ) . 2 In other words, the angular dependence of the probability density, |ψE (r)| , with shapes corresponding to the s, p, and d orbitals are shown in Figure 2.2.3. For those interested, further details can be found in Pettifor (1995) or Borg and Dines (1992). Figure 2.2.4 shows the energies of the valence electrons, defined as those that occupy the outermost shell, and principally involved in any bonding for all s and p levels. Unlike the hydrogen atom, where the energy levels (2.2.1) are determined by En ∝ n−2 , for atoms with multiple electrons, states with the same principal quantum number, n, but different orbital quantum number, l, have nondegenerate, or different, energy levels. This is because the presence of additional

Adapted from Callister and Rethwisch (2010).

OUP CORRECTED – FINAL

74

Atomic Structure and Spectra Table 2.2.1 Electronic structure of the atom, including the first ionization energy, for the first 36 elements. Principal Quantum Number, n Orbital Quantum Number, l Letter Designation of State Z 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

Symbol H He Li Be B C N O F Ne Na Mg Al Si P S Cl Ar K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr

Adapted from [1].

Element Hydrogen Helium Lithium Beryllium Boron Carbon Nitrogen Oxygen Fluorine Neon Sodium Magnesium Aluminum Silicon Phosphorus Sulfur Chlorine Argon Potassium Calcium Scandium Titanium Vanadium Chromium Manganese Iron Cobalt Nickel Copper Zinc Gallium Germanium Arsenic Selenium Bromine Krypton

1 0 1s Ionization Energy (eV) 13.60 24.58 5.39 9.32 8.30 11.26 14.54 13.61 17.42 21.56 5.14 7.64 5.98 8.15 10.55 10.36 13.01 15.76 4.34 6.11 6.56 6.83 6.74 6.76 7.43 7.90 7.86 7.63 7.72 9.39 6.00 7.88 9.81 9.75 11.84 14.00

2 0 2s

2 3 1 0 2p 3s

3 3 4 4 1 2 0 1 3p 3d 4s 4p

1 2 1 2 2 He 2 Core 2 2 2 2

Ne Core

1 2 3 4 5 6 1 2 2 2 2 2 2 2

Ar Core

1 2 3 4 5 6

1 2 3 4 5 6 7 8 10 10 10 10 10 10 10 10

1 2 2 2 2 2 2 2 2 2 1 2 2 2 2 2 2 2

1 2 3 4 5 6

OUP CORRECTED – FINAL

Atomic Structure 75 z

y x s z

z

z

y

y

y x

x

x px

py

pz z

z

y

y

dz2

dx2 – y2 z

z

y x

z

y

y

t2g

x

x dxy

eg

x

x

dyz

dxz

(more than one) electrons outside the nucleus, perturbs the potential, V (r), and it no longer shows a simple r –2 dependence. Three important observations can be made from Figure 2.2.4. First the 2s levels are all well below the 2p levels for the first-row elements, B through Ne. Second, both E p and E s depend linearly on the atomic number, Z, and not as Z 2 as expected from (2.2.1)—but why? Both of these observations can be explained in terms of the change in the potential due to the presence of other electrons. Finally, the energy difference, Ep −Es , decreases as one goes from the right to the left of the periodic table. This is significant because as the difference becomes small, it will

Figure 2.2.3 The angular component of the probability distribution corresponding to the s, p, and d orbitals. Adapted from Krishnan (2016).

OUP CORRECTED – FINAL

Atomic Structure and Spectra 0

Atomic energy level (eV)

76

–10 6p 5p 4p 3p

Ep

–20

6s

IA IIA

IIB IIIB IVB VB VIB VIIB

Li Na K Rb Cs

B C N O F Al Si P S Cl Zn Ga Ge As Se Br Cd In Sn Sb Te I Hg Ti Pb Bi P At

2p

5s

Es

2s

4s 3s

Be Mg Ca Sr Ba

Ne A Kr Xe Rn

–30 IA IIA

IIB IIIB IVB VB

VIIB

Figure 2.2.4 The valence s and p energy levels, E s and E p , respectively. Adapted from Pettifor (1995).

enable the valence s and p electrons to “hybridize” and form common sp orbitals (Fig. 2.2.5). As Chapter 3 shows, these hybridized orbitals and their bonding configurations provide signature differences in their unoccupied levels that can be probed with absorption spectra using either electron or X-ray incidence (see Problem 3.6). Lastly, it is helpful to briefly describe the concept of spin–orbit interaction or coupling. By changing the coordinate frame in an atom (Fig. 2.2.6), rather than considering the electron in orbit around the nucleus, we can keep the electron stationary and consider the nucleus in orbit around it with a radius equivalent to the Bohr radius. The orbiting nuclear charge, +Ze, produces a well-defined field/induction (think of a solenoid) at the position of the stationary electron. The intrinsic spin magnetic moment of the electron interacts with this field with its energy lower (higher) when they are both aligned parallel (antiparallel). Thus, we can have an effective coupling between the spin and the orbit of the electron.

OUP CORRECTED – FINAL

Atomic Structure 77 Z

(a)

(b)

Z

(c)

pz

ψ1

s

109.5º

109.5º

py

120°

ψ1

120°

px

109.5º

109.5º

Y

120°

Y

ψ1

X X

Figure 2.2.5 One s and three p orbitals shown in (a) can hybridize and form four sp3 orbitals (b) along the directions of a tetrahedron. These sp3 orbitals are key to the bonding in diamond. (c) If one s and two p orbitals hybridize, they form three sp2 orbitals, all lying in a single plane. These sp2 orbitals are key to the bonding in graphite. Adapted from Borg and Dienes (1992).

z

(b)

(a) Z

l +Ze

–e

l

B Y

y

+Ze

–e x

X Figure 2.2.6 The spin–orbit interaction can be explained by changing the frame of reference from the nucleus (a), to the orbiting electron (b). In (b) the orbiting nuclear charge, +Ze, generates an effective magnetic field/induction at the electron that can couple with its intrinsic spin with different energies depending on their relative orientation. Adapted from Krishnan (2016).

It is common practice to include this effective interaction between the spin and the magnetic field generated by the electron orbit, or spin–orbit coupling, by combining the spin and orbital contribution to the angular momentum for each electron (see, for example, Krishnan (2016), Sect. 2.8, for a more detailed discussion). Then, the appropriate set of quantum numbers are n = 1, 2, 3 . . . , l = 0, 1, 2, . . . n − 1, and two new quantum numbers, the total angular momentum quantum number j = l ± s, and mj = ± 1/2, ± 3/2, · · · ± j that gives its z-component, where both j and mj are, by definition, half integers. In this scenario, the shells are further split in energy to give a configuration such

OUP CORRECTED – FINAL

78

Atomic Structure and Spectra as 1s, 2s, 2p1/2 , 2p3/2 , 3s, 3p1/2 , 3p3/2 , 4s, 3d3/2 , 3d5/2 , .. . . . , where the additional subscript denotes the total angular momentum quantum number, j = l ± s, which accounts for the spin–orbit splitting. The difference in energies between the 2p1/2 and 2p3/2 levels, or the 2p1/2 − 2p3/2 splitting as it is called, is small for small Z (∼1 eV for Al), but increases as Z increases (∼23 eV for Zn, and ∼75 eV for Y). As we show in the next sections, this splitting is important and is included in the nomenclature of emission spectra. Table 2.2.2 summarizes the electron binding energies for all elements in their natural form. Note that the energies are given in eV with respect to the vacuum level for H2 , N2 , O2 , F2 , and Cl2 , and the rare gases, but relative to the Fermi level (Fig. 3.2.3) for metals, and relative to the top of the valence band for semiconductors (Fig. 3.7.1).

2.3 Atomic Spectra: Transitions, Emissions, and Secondary Processes Three important characterization or analysis techniques discussed in this chapter are based on atomic electron transitions. These include two-transition processes such as the emission of characteristic X-rays, which involves the creation of a core hole (first transition) and the decay of an outer electron (second transition) along with the emission of an X-ray photon. This process is the basis of the techniques of X-ray fluorescence (XRF; X-rays in, X-rays out; §2.5.2.1) and electron probe microanalysis (EPMA; electrons in, X-rays out; §2.5.2.2), or microanalysis in a TEM/SEM (energy dispersive X-ray spectrometry, EDXS; electrons in, X-rays out; §2.5.1.1, §9.4.3, and §10.6.1). It is also the basis of particle-induced X-ray emission (PIXE, high-energy particle in, X-rays out), discussed later in §5.4.5. Similarly, AES (electrons in, electrons out; §2.6.2), involves the creation of a core hole (first transition), followed by the Auger decay (second transition). Note that the emission of characteristic X-rays and Auger electrons are competing processes and the former (latter) is favored by high-Z (low-Z) elements (Fig. 1.4.5). An example of an analysis method consisting of a single transition is XPS (X-rays in and electrons out), which involves the simultaneous formation of a core hole and the emission of an energetic photoelectron. Of the many possibilities, only some transitions of electrons in atoms are observed; the rules governing these allowed transitions, as you may have surmised, depend on quantum mechanics, and they are discussed in the next section.

2.3.1 Dipole Selection Rules and Allowed Transitions of Electrons in Atoms We consider a radiative decay process consisting of a transition between two stationary states, i.e. the initial, higher energy state, ψ i , and the final, lower energy

Table 2.2.2 Electron binding energies (eV) of the elements in their natural form. Element

13.6 24.6* 54.7* 111.5* 188* 284.2* 409.9* 543.1* 696.7* 870.2* 1070.8† 1303.0† 1559.6 1839 2145.5 2472 2822.4 3205.9* 3608.4* 4038.5* 4492 4966 5465 5989 6539 7112 7709 8333 8979 9659 10367 11103 11867 12658

L1 2s

L2 2p1/2

L3 2p3/2

M1 3s

M2 3p1/2 M3 3p3/2 M4 3d3/2 M5 3d5/2 N1 4s N2 4p1/2 N3 4p3/2

21.7* 30.65 49.78 72.95 99.82 136* 163.6* 202* 250.6† 297.3* 349.7† 403.6* 460.2† 519.8† 583.8† 649.9† 719.9† 793.2† 870.0† 952.3† 1044.9* 1143.2† 1248.1*b 1359.1*b 1474.3*b

21.6* 30.81 49.50 72.55 99.42 135* 162.5* 200* 248.4* 294.6* 346.2† 398.7* 453.8† 512.1† 574.1† 638.7† 706.8† 778.1† 852.7† 932.7 1021.8* 1116.4† 1217.0*b 1323.6*b 1433.9*b

29.3* 34.8* 44.3† 51.1* 58.7† 66.3† 74.1† 82.3† 91.3† 101.0† 110.8† 122.5† 139.8* 159.5† 180.1* 204.7* 229.6*

15.9* 18.3* 25.4† 28.3* 32.6† 37.2† 42.2† 47.2† 52.7† 58.9† 68.0† 77.3† 91.4* 103.5† 124.9* 146.2* 166.5*

37.3* 41.6* 48.5* 63.5† 88.7 117.8 149.7*b 189* 230.9 270* 326.3* 378.6* 438.4† 498.0* 560.9† 626.7† 696.0† 769.1† 844.6† 925.1† 1008.6† 1096.7† 1196.2* 1299.0*b 1414.6*b 1527.0*b 1652.0*b

15.7* 18.3* 25.4† 28.3* 32.6† 37.2† 42.2† 47.2† 52.7† 59.9† 66.2† 75.1† 83.6* 100.0† 120.8* 141.2* 160.7*

10.2* 18.7† 29.8 41.7* 55.5*

OUP CORRECTED – FINAL

1H 2 He 3 Li 4 Be 5B 6C 7N 8O 9F 10 Ne 11 Na 12 Mg 13 Al 14 Si 15 P 16 S 17 Cl 18 Ar 19 K 20 Ca 21 Sc 22 Ti 23 V 24 Cr 25 Mn 26 Fe 27 Co 28 Ni 29 Cu 30 Zn 31 Ga 32 Ge 33 As 34 Se

K 1s

10.1* 18.7† 29.2 41.7* 54.6* continued

Element 35 Br 36 Kr 37 Rb 38 Sr 39 Y 40 Zr 41 Nb 42 Mo 43 Tc 44 Ru 45 Rh 46 Pd 47 Ag 48 Cd 49 In 50 Sn 51 Sb 52 Te 53 I 54 Xe 55 Cs 56 Ba 57 La 58 Ce 59 Pr 60 Nd 61 Pm 62 Sm 63 Eu 64 Gd

K 1s

L1 2s

L2 2p1/2

L3 2p3/2

M1 3s

13474 14326 15200 16105 17038 17993 18986 20000 21044 22117 23220 24350 25514 26711 27940 29200 30491 31814 33169 34561 35985 37441 38925 40443 41991 43569 45184 46834 48519 50239

1782* 1921 2065 2216 2373 2532 2698 2866 3043 3224 3412 3604 3806 4018 4238 4465 4698 4939 5188 5453 5714 5989 6266 6549 6835 7126 7428 7737 8052 8376

1596* 1730.9* 1864 2007 2156 2307 2465 2625 2793 2967 3146 3330 3524 3727 3938 4156 4380 4612 4852 5107 5359 5624 5891 6164 6440 6722 7013 7312 7617 7930

1550* 1678.4* 1804 1940 2080 2223 2371 2520 2677 2838 3004 3173 3351 3538 3730 3929 4132 4341 4557 4786 5012 5247 5483 5723 5964 6208 6459 6716 6977 7243

257* 292.8* 326.7* 358.7† 392.0*b 430.3† 466.6† 506.3† 544* 586.1* 628.1† 671.6† 719.0† 772.0† 827.2† 884.7† 946† 1006† 1072* 1148.7* 1211*b 1293*b 1362*b 1436*b 1511 1575 – 1723 1800 1881

M2 3p1/2 M3 3p3/2 M4 3d3/2 M5 3d5/2 189* 222.2* 248.7* 280.3† 310.6* 343.5† 376.1† 411.6† 447.6 483.5† 521.3† 559.9† 603.8† 652.6† 703.2† 756.5† 812.7† 870.8† 931* 1002.1* 1071* 1137*b 1209*b 1274*b 1337 1403 1471 1541 1614 1688

182* 214.4 239.1* 270.0† 298.8* 329.8† 360.6† 394.0† 417.7 461.4† 496.5† 532.3† 573.0† 618.4† 665.3† 714.6† 766.4† 820.0† 875* 940.6* 1003* 1063*b 1128*b 1187*b 1242 1297 1357 1420 1481 1544

70* 95.0* 113.0* 136.0† 157.7† 181.1† 205.0† 231.1† 257.6 284.2† 311.9† 340.5† 374.0† 411.9† 451.4† 493.2† 537.5† 583.4† 630.8 689.0* 740.5* 795.7† 853* 902.4* 948.3* 1003.3* 1052 1110.9* 1158.6* 1221.9*

69* 93.8* 112* 134.2† 155.8† 178.8† 202.3† 227.9† 253.9* 280.0† 307.2† 335.2† 368.3 405.2† 443.9† 484.9† 528.2† 573.0† 619.3 676.4* 726.6* 780.5* 836* 883.8* 928.8* 980.4* 1027 1083.4* 1127.5* 1189.6*

N1 4s

N2 4p1/2

27.5* 30.5* 38.9† 43.8* 50.6† 56.4† 63.2† 69.5* 75.0† 81.4*b 87.1*b 97.0† 109.8† 122.9† 137.1† 153.2† 169.4† 186* 213.2* 232.3* 253.5† 274.7* 291.0* 304.5 319.2* – 347.2* 360 378.6*

14.1* 16.3* 21.3 24.4* 28.5† 32.6† 37.6† 42.3* 46.3† 50.5† 55.7†a 63.7† 63.9†a 73.5†a 83.6†a 95.6†a 103.3†a 123* 146.7 172.4* 192 205.8 223.2 236.3 243.3 242 265.6 284 286

N3 4p3/2 14.1∗ 15.3∗ 20.1† 23.1∗ 27.1† 30.8† 35.5† 39.9∗ 43.2† 47.3† 50.9† 53.3† 63.9†a 73.5†a 83.6†a 95.6†a 103.3†a 123* 145.5* 161.3* 178.6† 196.0* 206.5* 217.6 224.6 242 247.4 257 271

OUP CORRECTED – FINAL

Table 2.2.2 Continued

65 Tb 66 Dy 67 Ho 68 Er 69 Tm 70 Yb

51996 53789 55618 57486 59390 61332

8708 9046 9394 9751 10116 10486

8252 8581 8918 9264 9617 9978

7514 7790 8071 8358 8648 8944

1968 2047 2128 2207 2307 2398

1768 1842 1923 2006 2090 2173

1611 1676 1741 1812 1885 1950

1276.9* 1333 1392 1453 1515 1576

N4 4d3/2

N5 4d5/2

N6 4f5/2

N7 4f7/2

O1 5s

O2 5p1/2

O3 5p3/2

48 Cd 49 In 50 Sn 51 Sb 52 Te 53 I 54 Xe 55 Cs 56 Ba 57 La 58 Ce 59 Pr 60 Nd 61 Pm 62 Sm 63 Eu 64 Gd 65 Tb 66 Dy 67 Ho 68 Er 69 Tm 70 Yb

11.7† 17.7† 24.9† 33.3† 41.9† 50.6 69.5* 79.8* 92.6† 105.3* 109* 115.1* 120.5* 120 129 133 – 150.5* 153.6* 160* 167.6* 175.5* 191.2*

10.7† 16.9† 23.9† 32.1† 40.4† 48.9 67.5* 77.5* 89.9† 102.5* – 115.1* 120.5* 120 129 127.7* 142.6* 150.5* 153.6* 160* 167.6* 175.5* 182.4*

– – – – 0.1 2.0 1.5 – 5.2 0 8.6* 7.7* 8.0* 8.6* – – 2.5*

– – – – 0.1 2.0 1.5 – 5.2 0 8.6* 2.4* 4.3* 5.2* 4.7* 4.6 1.3*

23.3* 22.7 30.3† 34.3* 37.8 37.4 37.5 – 37.4 32 36 45.6* 49.9* 49.3* 50.6* 54.7* 52.0*

13.4* 14.2* 17.0† 19.3* 19.8* 22.3 21.1 – 21.3 22 28 28.7* 26.3 30.8* 31.4* 31.8* 30.3*

12.1* 12.1* 14.8† 16.8* 17.0* 22.3 21.1 – 21.3 22 21 22.6* 26.3 24.1* 24.7* 25.0* 24.1*

O4 5d3/2

O5 5d5/2

396.0* 414.2* 432.4* 449.8* 470.9* 480.5* P1 6s

322.4* 333.5* 343.5 366.2 385.9* 388.7*

284.1* 293.2* 308.2* 320.2* 332.6* 339.7*

P2 6p1/2

P3 6p3/2

continued

OUP CORRECTED – FINAL

Element

1241.1* 1292.6* 1351 1409 1468 1528

OUP CORRECTED – FINAL

Table 2.2.2 Continued Element 71 Lu 72 Hf 73 Ta 74 W 75 Re 76 Os 77 Ir 78 Pt 79 Au 80 Hg 81 Tl 82 Pb 83 Bi 84 Po 85 At 86 Rn 87 Fr 88 Ra 89 Ac 90 Th 91 Pa 92 U

K 1s

L1 2s

L2 2p1/2

L3 2p3/2

M1 3s

M2 3p1/2

M3 3p3/2

M4 3d3/2

M5 3d5/2

63314 65351 67416 69525 71676 73871 76111 78395 80725 83102 85530 88005 90524 93105 95730 98404 101137 103922 106755 109651 112601 115606

10870 11271 11682 12100 12527 12963 13419 13880 14353 14839 15347 15861 16388 16939 17493 18049 18639 19237 19840 20472 21105 21757

10349 10739 11136 11544 11959 12385 12824 13273 13734 14209 14698 15200 15711 16244 16785 17337 17907 18484 19083 19693 20314 20948

9244 9561 9881 10207 10535 10871 11215 11564 11919 12284 12658 13035 13419 13814 14214 14619 15031 15444 15871 16300 16733 17166

2491 2601 2708 2820 2932 3049 3174 3296 3425 3562 3704 3851 3999 4149 4317 4482 4652 4822 5002 5182 5367 5548

2264 2365 2469 2575 2682 2792 2909 3027 3148 3279 3416 3554 3696 3854 4008 4159 4327 4490 4656 4830 5001 5182

2024 2108 2194 2281 2367 2457 2551 2645 2743 2847 2957 3066 3177 3302 3426 3538 3663 3792 3909 4046 4174 4303

1639 1716 1793 1872 1949 2031 2116 2202 2291 2385 2485 2586 2688 2798 2909 3022 3136 3248 3370 3491 3611 3728

1589 1662 1735 1809 1883 1960 2040 2122 2206 2295 2389 2484 2580 2683 2787 2892 3000 3105 3219 3332 3442 3552

N1 4s

N2 4p1/2

N3 4p3/2

506.8* 538* 563.4† 594.1† 625.4† 658.2† 691.1† 725.4† 762.1† 802.2† 846.2† 891.8† 939† 995* 1042* 1097* 1153* 1208* 1269* 1330* 1387* 1439*b

412.4* 438.2† 463.4† 490.4† 518.7† 549.1† 577.8† 609.1† 642.7† 680.2† 720.5† 761.9† 805.2† 851* 886* 929* 980* 1058 1080* 1168* 1224* 1271*b

359.2* 380.7† 400.9† 423.6† 446.8† 470.7† 495.8† 519.4† 546.3† 576.6† 609.5† 643.5† 673.8† 705* 740* 768* 810* 879* 890* 966.4† 1007* 1043†

Element

N4 4d3/2

N5 4d5/2

N6 4f5/2

N7 4f7/2

O1 5s

O2 5p1/2

O3 5d3/2

O4 5d3/2

O5 5d5/2

P1 6s

P2 6p1/2

P3 6p3/2

71 Lu 72 Hf 73 Ta 74 W 75 Re 76 Os 77 Ir 78 Pt 79 Au 80 Hg 81 Tl 82 Pb 83 Bi 84 Po 85 At 86 Rn 87 Fr 88 Ra 89 Ac 90 Th 91 Pa 92 U

206.1* 220.0† 237.9† 255.9† 273.9† 293.1† 311.9† 331.6† 353.2† 378.2† 405.7† 434.3† 464.0† 500* 533* 567* 603* 636* 675* 712.1† 743* 778.3†

196.3* 211.5† 226.4† 243.5† 260.5† 278.5† 296.3† 314.6† 335.1† 358.8† 385.0† 412.2† 440.1† 473* 507 541* 577* 603* 639* 675.2† 708* 736.2†

8.9* 15.9† 23.5† 33.6* 42.9* 53.4† 63.8† 74.5† 87.6† 104.0† 122.2† 141.7† 162.3† 184* 210* 238* 268* 299* 319* 342.4† 371* 388.2*

7.5* 14.2† 21.6† 31.4† 40.5* 50.7† 60.8† 71.2† 84.0 99.9† 117.8† 136.9† 157.0† 184* 210* 238* 268* 299* 319* 333.1† 360* 377.4†

57.3* 64.2† 69.7† 75.6† 83† 84* 95.2*b 101.7*b 107.2*b 127† 136.0*b 147*b 159.3*b 177* 195* 214* 234* 254* 272* 290*a 310* 321*ab

33.6* 38* 42.2* 45.3*b 45.6* 58* 63.0*b 65.3*b 74.2† 83.1† 94.6† 106.4† 119.0† 132* 148* 164* 182* 200* 215* 229*a 232* 257*ab

26.7* 29.9† 32.7† 36.8† 34.6*b 44.5† 48.0† 51.7† 57.2† 64.5†| 73.5† 83.3† 92.6† 104* 115* 127* 140* 153* 167* 182*a 232* 192*ab

9.6† 14.7† 20.7† 26.9† 31* 40* 48* 58* 68* 80* 92.5† 94* 102.8†

7.8† 12.5† 18.1† 23.8† 31* 40* 48* 58* 68* 80* 85.4† 94* 94.2†

26 34 44 – 41.4† – 43.9†

15 19 – 24.5† – 26.8†

15 19 – 16.6† – 16.8†

Adapted from X-ray Data Booklet, Published by the Center for X-ray optics and Advanced Light Source: LBNL/PUB-490 Rev-2 (2001).

OUP CORRECTED – FINAL

OUP CORRECTED – FINAL

Atomic Structure and Spectra state, ψ f , of the atom. During a transition from ψ i to ψ f , the average position of the   electron oscillates between these two states, with a frequency fif = Ei − Ef /h, given by the difference in their energies (Fig. 2.3.1a). This oscillation between these two states, referred to in quantum mechanics as a mixed state, goes on for a finite period of time, known as the lifetime of the transition. During this lifetime, the probability of finding the electron in the higher energy state, ψi , decreases gradually from unity to zero, while the probability of finding the electron in the lower energy state, ψf , increases gradually from zero to unity (Fig. 2.3.1b). The lifetime is sufficiently long that during the transition period the electron executes many millions of oscillations, back and forth, between ψi and ψf . Moreover, in quantum mechanical terms, the larger the number of oscillations in the period defined by the lifetime of the transition, the better defined is the wavelength of the emitted photon or wave packet (Fig. 1.3.5); this also translates into a narrow line width in the observed spectrum. In quantum mechanics the transition probability between states, ψi and ψf , is proportional to the square of the dipole matrix element −e ψi rψf dr = −erif

(2.3.1)

A detailed discussion of (2.3.1), which provides a method to compute the atomic transition probabilities between the two states, is beyond the scope of this book, but can be found in Quantum Mechanics textbooks, e.g. Rae (1992). However, suffice it to say that the transition from ψi to ψf occurs only when the matrix element gives a finite oscillation amplitude, rif (t). If the wave functions, ψi and ψf , are such that the integral, (2.3.1), is zero, the transition is not allowed as the oscillations are non-existent. By simple examination of the integral, one can see that r is an odd function of the coordinates (replacing r by – r changes its sign) (b)

Time

Probability

(a) Oscillation amplitude

84

1 Final State, ψf Initial State, ψi 0

Time

Figure 2.3.1 A radiative decay process consists of (a) the mixed atomic state oscillating between the initial, ψi , and final, ψf , states, with the probability (b) of the initial state decaying to zero during the lifetime—involving millions of such oscillations—of the transition. Adapted from Atwood (2000).

OUP CORRECTED – FINAL

Atomic Spectra: Transitions, Emissions, and Secondary Processes 85 and hence ψi and ψf should be of opposite parity (thus one of them should be odd, and the other even, in terms of the coordinates) to give a nonzero value of the integral. The parity of the wave functions alternates with increasing orbital quantum number, l, giving the first dipole selection rule for an observable or allowed transition in a single electron atom as l = ±1

(2.3.2a)

In addition, the total angular momentum quantum number, j = l ± s, must also satisfy j = 0, ±1

(2.3.2b)

A physical way to understand the selection rules is to consider the emitted photon as a particle with an angular momentum of unity. Thus, to conserve angular momentum and energy, transitions where angular momentum changes by one unit, i.e. l = ±1 and j = 0, ±1, are only allowed (Fig. 2.3.4).

2.3.2 Characteristic X-Ray Emissions and their Nomenclature Figure 2.3.2 shows the nomenclature used to describe X-rays generated from specific atomic transitions. To understand this nomenclature, consider the electron energy-level diagrams for two representative elements, sodium and molybdenum (Fig. 2.3.3). Here the energy of the entire atom is plotted, with zero indicating the neutral atom in its ground state. Any excitation, say the removal of a particular core electron, leaving behind a hole or a vacant quantum state, leads to an increase in energy of the atom (excited state). For example, the energy of 1070.8 eV associated with 1s level of Na (Fig. 2.3.3) indicates the energy of the excited state of the Na atom when

Kδ Kγ Kβ Kα

Mγ

Lα

Mβ Mα

Lβ Lγ Lδ

Figure 2.3.2 The nomenclature used to describe different X-ray lines arising from specific atomic transitions.

OUP CORRECTED – FINAL

Atomic Structure and Spectra 1s

1s

1071 eV K

20,002 eV K

Energy of atom (eV)

1000

Energy of atom (eV)

86

500

10,000

Sodium

Molybdenum

2s 2p

0

2s 2p 3s

63.4 e.V. 30.7 e.V. 5.14 e.V.

0

3s, p, d

2867 e.V. 2629 2524 508 e.V. 412 395 234 e.V. 230

2s 2p1/2 2p3/2

L

3s 3p1/2 3p3/2 3d3/2 3d5/2

M

Figure 2.3.3 Electron energy level diagram for Na and Mo. Excited states of the atom, upon the removal of specific core electrons, as indicated, are shown. Adapted from Sproul (1980).

the 1s electron from the K-shell (n = 1) has been removed. In this scenario, a 2p electron, following the dipole selection rule, (2.3.2a), can make a transition to 1s level, filling the K-shell, and then emitting an X-ray photon. This X-ray photon, called the Kα spectral line (or K-line), would have a characteristic energy of 1040.1 eV (= 1070.8 − 30.7). The selection rule would also allow a 3p → 1s transition (Kβ line), but this is not observed for Na as the 3p state is unoccupied. We shall illustrate these concepts further with the example of Mo.

OUP CORRECTED – FINAL

Atomic Spectra: Transitions, Emissions, and Secondary Processes 87 Example 2.3.1: Calculate the energies of the characteristic X-ray emissions for Mo. Solution: The binding energies for Mo (Fig. 2.3.3) are E1s = 20, 002 eV, and since the 2p levels are split due to spin–orbit coupling we have E2p3/2 = 2, 524 eV and E2p1/2 = 2, 629 eV. The resulting Kα X-ray line is also split into two: Kα1 = 20, 002 − 2524 = 17, 478 eV and Kα2 = 20, 002 − 2629 = 17, 373 eV. In addition, the Kβ lines are also split into two: Kβ1 = ls − 3p3/2 = 20002 − 395 = 19607 eV and Kβ3 = ls − 3p1/2 = 20002 − 412 = 19590 eV Alternatively, if the initial core hole were to be in the 2p (L –shell) level, we observe three X-ray lines Lα1 = 2p3/2 − 3d5/2 = 2524 − 230 = 2294 eV Lα2 = 2p3/2 − 3d3/2 = 2524 − 234 = 2290 eV and Lβ1 = 2p1/2 − 3d3/2 = 2629 − 234 = 2395 eV.

Kinetic energy (positive)

An alternative convention, which we favor in this book, is to plot the electron binding energies (negative) in all the shells in the multi-electron atom (Fig. 2.3.4).

0

Ekinetic = hf – EK,abs

n=∞ n=4

N

n=3

M

Binding energy (negative)

Lα Lβ

EL,abs

n=2

L hf Kβ

EK,abs

Figure 2.3.4 Energy levels for a multi-electron atom, showing the allowed transitions and the K-shell absorption edge.

Kα

K

n=1

Adapted from Atwood (2000).

OUP CORRECTED – FINAL

88

Atomic Structure and Spectra The transitions, following the allowed dipole selection rules, that show narrow emission lines, i.e. Kα (n = 2 → 1), Kβ (n = 3 → 1), Lα (n = 3 → 2), etc., are also indicated. Further, the energy, EK,abs , required to take an electron from the K-shell to the continuum (n = ∞) is called the absorption edge. If a photon (or electron) of energy greater than the absorption edge, hf > EK,abs , is incident, it can remove the K-shell electron beyond the continuum limit, to a state with positive kinetic energy. As an L-shell electron has a lower binding energy compared to the K-shell electron, if it is also lifted to the continuum by a similar photon, it would have a larger kinetic energy. These transitions are also shown in the same figure. Figure 2.3.5 shows a detailed energy level diagram for copper, for all the allowed transitions satisfying the selection rules, l = ±1 and j = 0, ±1, and including the spin–orbit coupling. The energy levels are not to scale. The reader is encouraged to study this figure carefully and understand each of the allowed n 4 4. .. 4

Absorption edges for copper (Z = 29):

l j 3 7/2 3. 5/2 . .. .. N 0 1/2

N7 4f7/2 : N4 4d3/2 : EN1, abs = 7.7 eV N1 4s Mα1

3 3 3 3 3

2 5/2 2 3/2 1 3/2 M 1 1/2 0 1/2

M5 M4 M3 M2 M1

3d5/2 3d3/2 3p3/2 3p1/2 3s

. . EM3, abs = 75 eV . EM1, abs = 123 eV

Lα1 Lα2 Lβ1

L3 2p3/2 EL3, abs = 933 eV L2 2p1/2 EL , abs = 952 eV EL21, abs = 1,097 eV L1 2s

2 1 3/2 2 1 1/2 L 2 0 1/2 Kα1 Kα2

Kβ1 Kβ3 Kγ3

1 0 1/2 K

K1s

EK, abs = 8,979 eV (1.381 Å)

Cu Kα1 = 8,048 eV (1.541Å) Cu Lα1 = 930 eV Cu Kα2 = 8,028 eV (1.544Å) Cu Lα2 = 930 eV Cu Lβ1 = 950 eV Cu Kβ1 = 8,905 eV Figure 2.3.5 Energy level diagram for copper showing all the transitions allowed by the dipole selection rules and including the spin–orbit coupling. Adapted from Atwood (2000).

OUP CORRECTED – FINAL

Atomic Spectra: Transitions, Emissions, and Secondary Processes 89 transitions and the nomenclature used to describe it, before attempting Problem 2.1. The characteristic X-ray photons generally have long mean free path lengths in solids (§2.4.3). If used as an exciting probe, they are also deeply penetrating (see §5.3). As a result, techniques that use X-rays as a probe (XRF) or signal (EPMA), give rise to bulk analytical tools. In contrast, the non-radiative Auger electron emission and photoelectron emission are both surface sensitive signals (Fig. 1.3.5) and are discussed in the next section. Table 2.3.1, adapted from [9], lists the energies of the principal X-ray emission lines for all elements.

Table 2.3.1 Energies (eV) of X-ray emission lines [9]. The corresponding wavelength in Å is given by λ = 12398/E, where E is in eV. Element Kα 1

Kα 2

Kβ 1

Lα 1

Lα 2

Lβ 1

3 Li 4 Be 5B 6C 7N 8O 9F 10 Ne 11 Na 12 Mg 13 Al 14 Si 15 P 16 S 17 Cl 18 Ar 19 K 20 Ca 21 Se 22 Ti 23 V 24 Cr 25 Mn 26 Fe 27 Co 28 Ni 29 Cu 30 Zn

848.6 1,040.98 1,253.60 1,486.27 1,739.38 2,012.7 2,306.64 2,620.78 2,955.63 3,311.1 3,688.09 4,086.1 4,504.86 4,944.64 5,405.509 5,887.65 6,390.84 6,915.30 7,460.89 8,027.83 8,615.78

1,071.1 1,302.2 1,557.45 1,835.94 2,139.1 2,464.04 2,815.6 3,190.5 3,589.6 4,012.7 4,460.5 4,931.81 5,427.29 5,946.71 6,490.45 7,057.98 7,649.43 8,264.66 8,905.29 9,572.0

341.3 395.4 452.2 511.3 572.8 637.4 705.0 776.2 851.5 929.7 1,011.7

341.3 395.4 452.2 511.3 572.8 637.4 705.0 776.2 851.5 929.7 1,011.7

344.9 399.6 458.4 519.2 582.8 648.8 718.5 791.4 868.8 949.8 1,034.7

54.3 108.5 183.3 277 392.4 524.9 676.8 848.6 1,040.98 1,253.60 1,486.70 1,739.98 2,013.7 2,307.84 2,622.39 2,957.70 3,313.8 3,691.68 4,090.6 4,510.84 4,952.20 5,414.72 5,898.75 6,403.84 6,930.32 7,478.15 8,047.78 8,638.86

Lβ 2 Lγ 1 Mα 1

OUP CORRECTED – FINAL

90

Atomic Structure and Spectra Table 2.3.1 Continued Element Kα 1

Kα 2

Kβ 1

Lα 1

Lα 2

Lβ 1

Lβ 2

Lγ 1

Mα 1

31 Ga 32 Ge 33 As 34 Se 35 Br 36 Kr 37 Rb 38 Sr 39 Y 40 Zr 41 Nb 42 Mo 43 Te 44 Ru 45 Rh 46 Pd 47 Ag 48 Cd 49 In 50 Sn 51 Sb 52 Te 53 I 54 Xe 55 Cs 56 Ba 57 La 58 Ce 59 Pr 60 Nd 61 Pm 62 Sm 63 Eu 64 Gd 65 Tb 66 Dy 67 Ho 68 Er 69 Tm 70 Yb

9,224.82 9,855.32 10,507.99 11,181.4 11,877.6 12,598 13,335.8 14,097.9 14,882.9 15,690.9 16,521.0 17,374.3 18,250.8 19,150.4 20,073.7 21,020.1 21,990.3 22,984.1 24,002.0 25,044.0 26,110.8 27,201.7 28,317.2 29,458 30,625.1 31,817.1 33,034.1 34,278.9 35,550.2 36,847.4 38,171.2 39,522.4 40,901.9 42,308.9 43,744.1 45,207.8 46,699.7 48,221.1 49,772.6 51,354.0

10,264.2 10,982.1 11,726.2 12,495.9 13,291.4 14,112 14,961.3 15,835.7 16,737.8 17,667.8 18,622.5 19,608.3 20,619 21,656.8 22,723.6 23,818.7 24,942.4 26,095.5 27,275.9 28,4860 29,725.6 30,995.7 32,294.7 33,624 34,986.9 36,378.2 37,801.0 39,257.3 40,748.2 42,271.3 43,826 45,413 47,037.9 48,697 50,382 52,119 53,877 55,681 57,517 59,370

1,097.92 1,188.00 1,282.0 1,379.10 1,480.43 1,586.0 1,694.13 1,806.56 1,922.56 2,042.36 2,165.89 2,293.16 2,424 2,558.55 2,696.74 2,838.61 2,984.31 3,133.73 3,286.94 3,443.98 3,604.72 3,769.33 3,937.65 4,109.9 4,286.5 4,466.26 4,650.97 4,840.2 5,033.7 5,230.4 5,432.5 5,636.1 5,845.7 6,057.2 6,272.8 6,495.2 6,719.8 6,948.7 7,179.9 7,415.6

1,097.92 1,188.00 1,282.0 1,379.10 1,480.43 1,586.0 1,692.56 1,804.74 1,920.47 2,039.9 2,163.0 2,289.85 2,420 2,554.31 2,692.05 2,833.29 2,978.21 3,126.91 3,279.29 3,435.42 3,595.32 3,758.8 3,926.04 — 4,272.2 4,450.90 4,634.23 4,823.0 5,013.5 5,207.7 5,407.8 5,609.0 5,816.6 6,025.0 6,238.0 6,457.7 6,679.5 6,905.0 7,133.1 7,367.3

1,124.8 1,218.5 1,317.0 1,419.23 1,525.90 1,636.6 1,752.17 1,871.72 1,995.84 2,124.4 2,257.4 2,394.81 2,538 2,683.23 2,834.41 2,990.22 3,150.94 3,316.57 3,487.21 3,662.80 3,843.57 4,029.58 4,220.72 — 4,619.8 4,827.53 5,042.1 5,262.2 5,488.9 5,721.6 5,961 6,205.1 6,456.4 6,713.2 6,978 7,247.7 7,525.3 7,810.9 8,101 8,401.8

2,219.4 2,367.0 2,518.3 2,674 2,836.0 3,001.3 3,171.79 3,347.81 3,528.12 3,713.81 3,904.86 4,100.78 4,301.7 4,507.5 — 4,935.9 5,156.5 5,383.5 5,613.4 5,850 6,089.4 6,339 6,586 6,843.2 7,102.8 7,366.7 7,635.7 7,911 8,189.0 8,468 8,758.8

2,302.7 2,461.8 2,623.5 2,792 2,964.5 3,143.8 3,328.7 3,519.59 3,716.86 3,920.81 4,131.12 4,347.79 4,570.9 4,800.9 — 5,280.4 5,531.1 5,788.5 6,052 6,322.1 6,602.1 6,892 7,178 7,480.3 7,785.8 8,102 8,418.8 8,747 9,089 9,426 9,780.1

833 883 929 978 — 1,081 1,131 1,185 1,240 1,293 1,348 1,406 1,462 1,521.4

9,251.74 9,886.42 10,543.72 11,222.4 11,924.2 12,649 13,395.3 14,165 14,958.4 15,775.1 16,615.1 17,479.34 18,367.1 19,279.2 20,216.1 21,177.1 22,162.92 23,173.6 24,209.7 25,271.3 26,359.1 27,472.3 28,612.0 29,779 30,972.8 32,193.6 33,441.8 34,719.7 36,026.3 37,361.0 38,724.7 40,118.1 41,542.2 42,996.2 44,481.6 45,998.4 47,546.7 49,127.7 50,741.6 52,388.9

OUP CORRECTED – FINAL

Atomic Spectra: Transitions, Emissions, and Secondary Processes 91 Table 2.3.1 Continued Element Kα 1

Kα 2

Kβ 1

Lα 1

Lα 2

Lβ 1

Lβ 2

Lγ 1

Mα 1

71 Lu 72 Hf 73 Ta 74 W 75 Re 76 Os 77 Ir 78 Pt 79 Au 80 Hg 81 Tl 82 Pb 83 Bi 84 Po 85 At 86 Rn 87 Fr 88 Ra 89 Ac 90 Th 91 Pa 92 U 93 Np 94 Pu 95 Am

52,965.0 54,611.4 56,277 57,981.7 59,717.9 61,486.7 63,286.7 65,112 66,989.5 68,895 70,831.9 72,804.2 74,814.8 76,862 78,950 81,070 83,230 85,430 87,670 89,953 92,287 94,665 — — —

61,283 63,234 65,223 67,244.3 69,310 71,413 73,560.8 75,748 77,984 80,253 82,576 84,936 87,343 89,800 92,300 94,870 97,470 100,130 102,850 105,609 108,427 111,300 — — —

7,655.5 7,899.0 8,146.1 8,397.6 8,652.5 8,911.7 9,175.1 9,442.3 9,713.3 9,988.8 10,268.5 10,551.5 10,838.8 11,130.8 11,426.8 11,727.0 12,031.3 12,339.7 12,652.0 12,968.7 13,290.7 13,614.7 13,944.1 14,278.6 14,617.2

7,604.9 7,844.6 8,087.9 8,335.2 8,586.2 8,841.0 9,099.5 9,361.8 9,628.0 9,897.6 10,172.8 10,449.5 10,730.91 11,015.8 11,304.8 11,597.9 11,895.0 12,196.2 12,500.8 12,809.6 13,122.2 13,438.8 13,759.7 14,084.2 14,411.9

8,709.0 9,022.7 9,343.1 9,672.35 10,010.0 10,355.3 10,708.3 11,070.7 11,442.3 11,822.6 12,213.3 12,613.7 13,023.5 13,447 13,876 14,316 14,770 15,235.8 15,713 16,202.2 16,702 17,220.0 17,750.2 18,293.7 18,852.0

9,048.9 9,347.3 9,651.8 9,961.5 10,275.2 10,598.5 10,920.3 11,250.5 11,584.7 11,924.1 12,271.5 12,622.6 12,979.9 13,340.4 — — 14,450 14,841.4 — 15,623.7 16,024 16,428.3 16,840.0 17,255.3 17,676.5

10,143.4 10,515.8 10,895.2 11,285.9 11,685.4 12,095.3 12,512.6 12,942.0 13,381.7 13,830.1 14,291.5 14,764.4 15,247.7 15,744 16,251 16,770 17,303 17,849 18,408 18,982.5 19,568 20,167.1 20,784.8 21,417.3 22,065.2

1,581.3 1,644.6 1,710 1,775.4 1,842.5 1,910.2 1,979.9 2,050.5 2,122.9 2,195.3 2,270.6 2,345.5 2,422.6 — — — — — — 2,996.1 3,082.3 3,170.8 — — —

54,069.8 55,790.2 57,532 59,318.24 61,140.3 63,000.5 64,895.6 66,832 68,803.7 70,819 72,871.5 74,969.4 77,107.9 79,290 81,520 83,780 86,100 88,470 90,884 93,350 95,868 98,439 — — —

Adapted from X-ray Data Booklet, Published by the Center for X-ray optics and Advanced Light Source: LBNL/PUB-490 Rev-2 (2001).

2.3.3

Non-Radiative Auger Electron Emission

Auger electron emission (§1.4.1) was discovered independently by Lise Meitner (in 1922) and Pierre Auger (in 1923), but named only after the latter! When an atom is ionized with the formation of a core hole by either an incident photon or electron of sufficient energy, the excited atom can lower its energy by filling the core hole with an electron from an outer level, along with the emission of energy. The emitted energy can be in the form of characteristic X-ray photons, following dipole selection rules (§2.3.2). Alternatively, the de-excitation process can be accompanied by a nonradiative emission of electrons (Fig. 2.3.6), known as Auger emissions. Note that in the process of de-excitation and Auger electron emission, the atom is left in its final state with two vacancies, or holes. Further, in the case of

OUP CORRECTED – FINAL

92

Atomic Structure and Spectra

e–

e–

e–

M 3s

M1

L3 L2 L1

2p3/2 2p1/2 2s

12

2

1 2

1

1s

K (a) KL1L1

(b) L1M1M1

(c) L1L2M1 Coster-Kronig

Figure 2.3.6 Schematic representation of three different two-electron (ABC) de-excitation processes. (a) The KL 1 L1 transition, starting with a core hole in the K-shell (step A), which is filled with an L 1 electron (step B); the other L 1 electron is ejected to the continuum (step C). Table 2.3.2 gives the energies of all KLL lines. (b) A similar L1 M1 M Auger process but with L-shell core hole. (c) The L1 L2 M1 line, known as a Coster–Kronig process. Note that in each case, eventually two holes, as indicated by numbers 1 and 2, are formed. Auger electron emission, the dipole selection rules are not followed. Figure 2.3.6 shows three cases, i.e. KL 1 L1 , L1 M1 M1 , and L1 L2 M1 , of Auger emission. The L1 L2 M1 line, specifically known as a Coster–Kronig process, where the initial and final vacancies are in the same shell (but not in the same sub-shell), tends to have a higher probability of occurring than the other Auger lines. The kinetic energy of a specific Auger electron, in the general case involving three levels, A, B, and C, is given by EABC = EA − EB − EC

(2.3.3)

where, EA , EB , and EC are one-electron binding energies, and EABC is the kinetic energy of the emitted Auger electron. In the case of the KL 1 L1 line shown in Figure 2.3.6, (2.3.3) can be specifically written as EKL1 L1 = EB,K − EB,L1 − EB,L1

(2.3.3a)

Note that, typically, one core level (E A ) that is characteristic of the atomic species and remains unchanged in any bonding, is always involved in the Auger emission. Hence this is a form of element-specific, core-level spectroscopy. This relationship, (2.3.3), though reasonable, is not a very accurate energy description of the Auger electron, for it does not take into account that the true energy is the

OUP CORRECTED – FINAL

Atomic Spectra: Transitions, Emissions, and Secondary Processes 93 difference between the binding energies of a one-hole and two-hole state in the atom. Thus, for a proper description of the kinetic energy of the Auger electron, the common practice is to rewrite (2.3.3) as EABC = EA − EB − EC − U

(2.3.4)

where,U = H − P, with H being the hole-hole interaction energy and P takes care of the extra atomic screening or relaxation effects in the presence of the holes. Typically, U, is a parameter that is experimentally determined. However, the kinetic energy of the Auger electron for an element, Z, can also be empirically written as Z Z = EAZ − EBZ − EC − EABC

 1  Z+1 Z EC − EC + EBZ+1 − EBZ 2

(2.3.4a)

Recall that the emission of characteristic X-rays and Auger electrons are complementary processes (Fig. 1.4.5). The yields of X-rays (W X ) and Auger electrons (W A ) are also given semi-empirically as a function of atomic number, Z:  4 WX = −a + bZ − cZ 3 WA

(2.3.5)

where a = 6.4 × 10−2 , b = 3.4 × 10−2 , and c = 1.03 × 10−6 . A typical Auger electron spectrum (Fig. 2.6.6) is a plot of the intensity versus kinetic energy and further details of this spectroscopy technique for surface chemical analysis are discussed in §2.6.2. Example 2.3.2: (a) Estimate the energy of the KL1 L2 Auger electrons for Ni. (b) Also calculate the fraction of Auger electron yield for Ni. Solution: (a) We shall apply (2.3.4a) and determine this empirically. The electron binding energies (keV) can be obtained from Table 2.2.2. They are Ni = 8.333, E Ni = 1.008, E Ni = 0.872, E Cu = 0.951, E Cu = 1.096 EK L1 L2 L2 L2 Then, from (2.3.4a), we get Ni Ni − E Ni − E Ni − EKL = EK L1 L2 1 L2

1 2



ELCu − ELNi2 + ELCu − ELNi1 2 1



= 8.333 − 1.008 − 0.872 − 12 (0.951 − 0.872 + 1.096 − 1.008) = 6.369 keV The observed value (Table 2.3.2) 6.384 keV, is in reasonably good agreement.

Table 2.3.2 KLL Auger lines (eV).

1S

C N O F Ne Na Mg Al Si P S Cl A K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr Rb Sr

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38

2s1 2p5 1P

0

1

KL1 L1

KL1 L2

243 356 474 610 761 928 105 301 516 742 982 249 527 815 122 456 799 168 557 956 374 808 264 732 214 712 216 749 283 840 412 995 595

252 362 486 627 781 952 135 336 554 784 034 305 586 881 195 533 886 259 651 056 480 923 384 861 348 852 365 903 447 014 594 186 795

1 1 1 1 1 2 2 2 3 3 3 4 4 4 5 5 6 6 7 7 8 8 9 9 10 10 11

1 1 1 1 2 2 2 2 3 3 3 4 4 5 5 5 6 6 7 7 8 8 9 10 10 11 11

3P

3P

0

KL1 L2 258 369 495 638 794 967 1 151 1 354 1 574 1 805 2 057 2 329 2 612 2 909 3 224 3 563 3916 4 290 4 683 5 089 5 514 5 957 6 419 6 896 7 384 7 888 8 401 8 939 9 483 10 049 10 630 11 221 11 830

2s2 2p4 3P

1

1S

2

0

KL1 L3

KL1 L3

KL2 L2

258 369 495 638 794 967 151 354 574 806 058 330 613 910 225 504 919 293 687 094 519 964 426 905 394 900 416 957 504 074 658 255 870

258 369 495 638 794 967 151 354 575 806 059 331 614 912 227 567 922 298 692 100 527 972 436 916 407 915 433 975 524 096 682 280 897

265 373 504 050 808 984 172 379 602 835 096 370 656 959 279 622 985 362 757 169 598 049 514 000 493 000 523 063 616 189 777 376 992

1 1 1 1 2 2 2 2 3 3 3 4 4 5 5 5 6 6 7 7 8 8 9 10 10 11 11

1 1 1 1 2 2 2 2 3 3 3 4 4 5 5 5 6 6 7 7 8 8 9 10 10 11 11

1 1 1 1 2 2 2 2 3 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 11 11

1D

3P

KL2 L3

KL3 L3

2

266 375 507 654 813 989 1 179 1 387 1 611 1 845 2 107 2 382 2 669 2 973 3 294 3 638 4 002 4 381 4 778 5191 6 622 6 075 6 542 7 030 7 526 8 037 8 563 9 107 9 665 10 244 10 837 11 442 12 066

3P

0

267 377 509 657 816 993 1 183 1 392 1 616 1 851 2 114 2 389 2 677 2 981 3 303 3 647 4011 4 391 4 788 5 202 5 634 6 088 6 556 7 045 7 543 8 057 8 586 9 133 9 695 10 279 10 877 11 487 12 118

2

KL3 L3

1 1 1 1 2 2 2 2 3 3 4 4 4 5 5 6 6 7 7 8 8 9 9 10 10 11 12

267 377 509 657 816 993 183 392 617 852 115 391 679 984 306 651 016 397 795 211 644 099 568 059 558 073 603 152 715 300 899 511 143

OUP CORRECTED – FINAL

2s0 2p6

39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 68 59 60 61 62 63 64 65 66 67 68 69 70 71 72

12 213 12 851 13 505 14 179 14 867 15 574 16 298 17 040 17 797 18 568 19 354 20 157 20 977 21 814 22 668 23 527 24 426 25 330 26 251 27 201 28 171 29 163 30 170 31 199 32 247 33 315 34 402 35 512 36 640 37 788 38 958 40 151 41 361 42 589

12 422 13 069 13 731 14 414 15 111 15 827 16 560 17 312 18 078 18 857 19 653 20 465 21 295 22 142 23 006 23 879 24 783 25 697 26 631 27 590 28 572 29 574 30 592 31 631 32 690 33 769 34 868 35 988 37 127 38 287 39 469 40 674 41 897 43 137

12 457 13 104 13 766 14 449 15 146 15 862 16 595 17 347 18 113 18 892 19 688 20 501 21 331 22 179 23 043 23 916 24 820 25 735 26 669 27 628 28 610 29 612 30 631 31 671 32 730 33 809 34 909 36 029 37 169 38 329 39 512 40 716 41 940 43 181

12 503 13 157 13 827 14 519 15 226 15 952 16 697 17 462 18 242 19 037 19 849 20 680 21 529 22 398 23 284 24 182 25 111 26 053 27 018 28 009 29 024 30 063 31 120 32 200 33 303 34 429 35 576 36 749 37 944 39 162 40 406 41 675 42 967 44 280

12 532 13 188 13 860 14 554 15 263 15 991 16 738 17 504 18 286 19 082 19 896 20 728 21 579 22 449 23 338 24 237 25 167 26 111 27 077 28 069 29 086 30 126 31 184 32 266 33 370 34 497 35 646 36 820 38 016 39 236 40 481 41 752 43 045 44 359

12 626 13 279 13 948 14 639 15 343 16 066 16 806 17 505 18 339 19 125 19 930 20 750 21 588 22 444 23 316 24 201 25 109 26 033 26 978 27 945 28 936 29 947 30 976 32 024 33 092 34 182 35 291 36 421 37 570 38 740 39 934 42 192 42 383 43 635

12 708 13 370 14 049 14 750 15 466 16 202 16 956 17 729 18 519 19 322 20 144 20 984 21 844 22 722 23 618 24 530 25 463 26 416 27 393 28 393 29 420 30 468 31 537 32 627 33 740 34 877 36 036 37 220 38 425 39 655 40 911 41 149 43 496 44 821

12 767 13 437 14 125 14 836 15 563 16 310 17 077 17 864 18 668 19 488 20 327 21 185 22 066 22 965 23 884 24 822 25 781 26 762 27 769 28 802 29 863 30 948 32 056 33 186 34 345 35 528 36 736 37 972 39 234 40 522 41 840 43 186 44 559 45 967

12 793 13 464 14 153 14 865 15 593 16 341 17 109 17 897 18 702 19 523 20 364 21 223 22 104 23 005 23 925 24 863 25 823 26 805 27 813 28 847 29 909 30 905 32 104 33 235 34 395 35 579 36 788 38 025 39 287 40 576 41 895 43 242 44 617 46 015

OUP CORRECTED – FINAL

Y Zr Nb Mo Te Ru Rh Pd Ag Cd In Sn Sb Te I Xe Cs Ba La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf

Table 2.3.2 Continued

Ta W Re Os Fr Pt Au Hg Tl Pb Bi Po At Rn Fr Ra Ac Th Pu U Np Pa Am Cm Bk Cf Es Fm Md No Lr Ku

73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104

2s1 2p5

2s2 2p4

1S 0 KL1 L1

1P 1 KL1 L2

3P 0 KL1 L2

3P 1 KL1 L3

3P 2 KL1 L3

1S 0 KL2 L2

1D 2 KL2 L3

3P 0 KL3 L3

43 831 45 097 46 385 47 690 49 022 50 375 51 752 53 149 54 554 55 992 57 451 58 918 60 427 61 980 63 523 65 103 66 720 68 341 70 016 71 704 73 437 75 204 77 060 78 867 80 594 83 286 85 219 87 205 89 221 91 267 93 373 95 518

44 391 45 671 46 972 48 291 49 636 51 003 52 393 53 802 55 227 56 677 58 155 59 640 61 163 62 720 64 286 66 887 67 509 69 153 70 842 72 550 74 297 76 080 77 930 79 590 81 528 84 187 86 146 88 144 90 192 92 260 94 388 96 555

44 436 46 715 47 018 48 337 49 682 51 050 52 440 53 849 55 275 56 726 58 205 59 690 61 213 62 771 64 337 65 939 67 562 69 207 70 896 72 604 74 351 76 135 77 985 79 646 81 585 34 245 86 204 88 203 90 251 92 320 94 448 96 615

45 611 46 971 48 357 49 767 51 205 52 672 54 167 55 685 57 225 58 799 60 402 62 026 63 689 65 392 67 114 68 879 70 673 72 498 74 373 76 280 78 236 80 237 82 317 84 386 86 408 89 453 91 701 93 998 96 356 98 763 101 250 103 796

45 691 47 053 48 440 49 851 51 291 52 759 54 255 55 774 57 316 58 891 60 495 62 120 63 784 65 489 67 212 68 978 70 774 72 600 74 476 76 384 78 342 80 344 82 425 84 495 86 518 89 565 91 814 94 113 96 471 98 880 101 368 103 915

44 900 46 193 47 507 48 839 50 195 51 575 52 978 54 397 55 840 57 302 58 799 60 299 61 836 63 397 64 983 66 604 68 232 69 898 71 599 73 327 75 085 76 884 78 727 80 240 82 388 85 017 86 997 89 006 91 085 93 173 95 322 97 510

46 164 47 538 48 938 50 361 51 812 53 292 54 801 56 330 57 890 59 476 61 098 62 739 64 416 66 124 67 868 69 654 71 453 73 302 75 190 77 116 79 086 81 103 83 177 85 099 87 331 90 348 92 617 94 926 97 315 99 744 102 252 104 820

47 377 48 831 50 315 51 830 53 375 54 954 56 568 58 206 59 882 61 591 63 338 65 118 66 935 68 789 70 690 72 640 74 611 76 040 78 714 80 839 83 019 85 254 87 558 89 888 92 204 95 607 98 165 100 774 103 472 106 240 109 108 112 055

Adapted from Siegbahn et al. (1967).

3P 2 KL3 L3

47 436 48 891 50 376 51 892 53 437 55 017 56 633 58 272 59 948 61 658 63 406 65 187 67 005 68 860 70 762 72 712 74 684 76 714 78 789 80 916 83 096 86 332 87 837 89 968 92 284 95 688 98 248 100 857 103 556 106 325 109 194 112 142

OUP CORRECTED – FINAL

2s0 2p6

OUP CORRECTED – FINAL

X-Rays as Probes: Generation and Transmission of X-Rays 97

(b) To calculate the fractional Auger electron yield for Ni (Z = 28) we apply (2.3.5) and get WX WA

 4 = −6.4 × 10−2 + 3.4 × 10−2 × 28 − 1.03 × 10−6 × 283 = 0.8654

But WX + WA = 1. Hence WA = 0.5361 or 53.6%.

2.3.4 Electron Photoemission For all practical purposes, the emission of electrons from atoms upon photon incidence is a very simple process involving a single transition (Fig. 1.4.2). An electron with a specific binding energy, E B , absorbs a photon of energy, hf, and then the same electron emerges from the atom with a kinetic energy, EK = hf −EB . Thus, the observed energy distribution of all photo-emitted electrons is simply related to the binding energies of all electrons in the atom by a simple shift of the energy scale by hf. In a solid, this simple picture is valid only if the probability of the photon being absorbed by all the electronic states in the solid is the same; unfortunately, this assumption is seldom true. Further in a solid, the choice of incident photon is simple; as long as hf also exceeds the work function, , or the energy required for the electron to escape from the surface of the material, it can be used. In practice, soft X-ray Al Kα (1486.7 eV) and Mg Kα (1253.6 eV) photons, readily available using laboratory sources, are employed. Strictly speaking, the work function of the specimen and spectrometer should be accounted for in the energy balance. If the specimen and spectrometer are in electrical contact and thus in thermodynamic equilibrium, their Fermi levels are equal and then the measured kinetic energy (Fig. 2.3.7) EK,meas = hf −EB − spect . XPS is introduced further in §2.6.3 and discussed in detail in §3.8.

2.4 X-Rays as Probes: Generation and Transmission of X-Rays In principle, there are two ways of generating X-rays using high-energy electrons to bombard a metal surface such as in an X-ray tube (Fig. 2.4.2). The electrons ionize the atoms on the surface of the target metal, and in the subsequent relaxation process, following dipole selection rules, produce characteristic X-ray photons (§2.3.1). In addition, all charged particles emit electromagnetic radiation when they are accelerated or decelerated. Thus, the high-energy electrons, on deceleration in the target metal, generate a continuous spectrum of radiation known as breaking radiation, or bremsstrahlung in German. A typical X-ray

OUP CORRECTED – FINAL

98

Atomic Structure and Spectra e– EK, meas Vacuum Level

EK Vacuum Level Figure 2.3.7 Schematic of the energy levels of the specimen and spectrometer in thermodynamic equilibrium. The measured kinetic energy of the photoemitted electron is now EK,meas = hf − EB − spect , where all energy terms are marked in the figure. Adapted from Feldman and Mayer (1986).

hf

spect s Fermi Level

Fermi Level

EB

Specimen

Spectrometer

spectrum Fig. 2.4.1 produced by the bombardment of a metal (Mo) target is a combination of these two processes. A continuous band of bremsstrahlung X-rays, starting with a ridge corresponding to the energy of the incident electrons, λSWL , and continuing through much larger wavelengths (lower energies) is observed as the background. Superimposed on this background are the characteristic X-ray peaks. Accelerating charged particles (electrons or positrons), such as those that are made to travel in curved trajectories using the Lorentz force (5.2.10) generated by a magnetic field, also emit X-rays of much higher intensities. In addition, if the charged particles were to move at relativistic speeds, as in a synchrotron, the emitted radiation is directed tangentially outward with respect to the particle motion in the shape of a narrow cone. Details of X-ray generation in laboratory-scale X-ray tubes and synchrotron sources, with access available to the latter as national user facilities in many countries, are discussed later.

2.4.1 Laboratory Sources and Methods of X-Ray Generation Figure 2.4.2 shows a schematic drawing of an X-ray tube. It contains a filament heated to high temperatures as the source of electrons, a set of two electrodes— one of which is the target (anode)—across which a high voltage, V, is applied to accelerate the electrons rapidly towards the water-cooled target (anode). All these components are sealed in a vacuum tube. X-rays produced at the point of impact are guided out of the tube through windows. Since only 1% of the electron

OUP CORRECTED – FINAL

X-Rays as Probes: Generation and Transmission of X-Rays 99

Kα1

X-ray intensity (relative units)

5

4

Kα Characterstic radiation Kα2

Continuous radiation

25 kV 3

is

Kβ 20 2 15

0.70

0.71

0.72

1 10 SWL 0

0

5 1.0

2.0

3

Wavelength (Å)

Figure 2.4.1 X-ray intensity as a function of wavelength for a Mo X-ray tube, as a function of applied voltage (kV). The natural line widths (Kα 1 ∼2 eV, λFWHM ∼ 0.001Å) are not to scale. Inset shows details of the Kα emission split into Kα 1 and Kα 2 emissions due to spin–orbit coupling. Adapted from Cullity (1977).

energy is converted to X-rays—the remaining 99% is converted to heat—extensive cooling of the target is required. Figure 2.4.1 shows how the intensity of the X-rays varies with wavelength. It is zero up to a certain wavelength, λSWL , called the short wavelength limit (SWL), and corresponds to the energy of the incident electrons, and then increases rapidly to a maximum before decreasing gradually as the wavelength increases. Electrons in the tube that are stopped in one impact by the target, generate photons with the maximum energy (or, minimum wavelength, λ) by transferring all their energy   (eV ) to the photon hf max . Thus eV = hf max = hc/λmin , or λmin = λSWL =

hc 12.39 × 103 = (Å) eV V

(2.4.1)

OUP CORRECTED – FINAL

100

Atomic Structure and Spectra High voltage

Target

Filament

Cooling Electrons Figure 2.4.2 A schematic drawing of an X-ray tube showing the principal components.

X-rays

High vacuum

Note that λSWL decreases as the high voltage in the tube, V, increases. If the electron does not lose all its energy in a single impact, but continues to move in the metal target with reduced velocity after the impact, then the emitted photon has energy less than hf max , and its wavelength is λ > λSWL . The bremsstrahlung reflects the total contribution of all such electrons that undergo partial transfer of energy to the photons. As the applied voltage in the tube is increased, both the average energy of the photons and the total number of photons increase, explaining the observed variation with voltage in Figure 2.4.1. The total intensity, I B , of the X-ray background, given by the area under any one curve, is a function of the atomic number, Z, of the target, the voltage, V, of the tube, and the current, i. It can be written as IB ∝ iZV m , where the exponent, m∼2. Superimposed on the background are the characteristic X-ray lines. For, Mo, with a binding energy, E B ∼20.002 keV, for the 1s electron, a K-line only appears if the excitation voltage, V, of the electron is larger than this value; thus, these characteristic emission lines do not appear in the lower energy curves in Figure 2.4.1. The intensity of the K-line, IK−line , is given by IK−line ∝ i(V − EB )n , where 1 < n < 2. The narrow width of the K-line, defined by its full width at half maximum, λFWHM > γ I

e– λ

hf Figure 2.4.5 Magnets used for synchrotron radiation and the X-ray intensities they produce as a function of energy: (a) bend magnets, (b) undulators with small angular excursions, and (c) wigglers with angular excursions larger than 1/γ . Adapted from Atwood (1999).

10 11

https://als.lbl.gov/ https://www.aps.anl.gov/

magnetic field excursions are larger than 1/γ ; now, the accelerations are stronger and the wiggler produces a higher photon flux and energy peak, with a broad radiation spectrum, similar to a bend magnet. Last but not least, bending magnet radiation is linearly polarized in the horizontal plane of acceleration. Above and below this plane, the radiation is elliptically polarized—a feature that has also been used to probe magnetic properties and image magnetic domains. See Figure 1.3.4 or Figure 6.9.4 for a description of elliptical polarization. Historically, high-energy physicists considered synchrotron radiation as a nuisance for it was a source of energy loss in electron storage rings. At first, parasitic ports were built in such storage rings for scientific experiments using X-ray photons. Subsequent second-generation synchrotron sources were dedicated facilities for X-ray photons built with bend magnets. Third-generation synchrotrons, of which the ALS10 (Fig. 2.4.6) and APS11 are good examples, have many straight sections optimized for the insertion of undulators and wigglers to produce soft and hard X-rays, respectively. The excellent text by Atwood (1999) has further details of synchrotron radiation and its wide range of applications.

OUP CORRECTED – FINAL

X-Rays as Signals: Core-Level Spectroscopy with X-Rays 107 (a)

(b) ND

XRD NR

SANS

XMCT

Advanced Light Sources based techniques

SAXS

SRCD

PIXE XRF

XPS XAFS

STXM

Figure 2.4.6 The Advanced Light Source (ALS) at the Lawrence Berkeley National Laboratory in Berkeley, CA. (a) The interior of the synchrotron, showing the beam lines in the periphery. (b) Schematic illustration of the different beam lines available and the wide range of X-ray based characterization methods implemented at the ALS. See https://als.lbl.gov/ for specific details of the ring and different end-stations.

2.5 X-Rays as Signals: Core-Level Spectroscopy with X-Rays An inner-shell vacancy in the atom can be created with X-ray photons, highenergy electrons, or accelerated particles. The core-shell vacancy can, in turn, be occupied by an outer shell electron and accompanied by the spontaneous emission of an element-specific X-ray photon characteristic of the atom. The core-levels are unperturbed by bonding or the immediate environment of the atom (§3.3), and hence detection of these characteristic X-ray lines helps unequivocally identify the constituent atoms in the material. Moreover, their intensities can be interpreted in terms of the composition of the specimen. Such microanalysis using characteristic X-ray emission can be carried out with X-ray, electron, or particle incidence (probes). Before we discuss these methods—XRF, EPMA, and PIXE—we must first understand how we detect X-ray photons and measure their intensities.

2.5.1 Instrumentation for Detecting X-Rays X-ray photons or wave packets are characterized by both their energy and wavelength. For example, Cu Kα1 radiation corresponds to a wavelength of 1.541 Å or an energy of 8.048 keV. Thus, we can detect them and measure their intensities using either wavelength or energy dispersion methods. 2.5.1.1

Wave Length Dispersion Spectrometer (WDS)

Figure 2.5.1a shows a WDS, which consists of an analyzing/diffracting crystal of finite size, typically 25x10 mm, that diffracts X-rays of selected wavelength by

OUP CORRECTED – FINAL

108

Atomic Structure and Spectra Incident Probe

Incident Probe

Diffracting crystal Crystal lattice spacing (d) θ

λ

Specimen

Crystal

Focusing circle Diffracted X-rays

X-rays detector

Take off angle Specimen

(a)

(b)

Detector

Figure 2.5.1 A wavelength dispersive spectrometer. (a) Schematic of the apparatus illustrating the diffracting crystal and detector geometry to select a narrow wavelength range that is diffracted. (b) As the take-off angle of the X-ray photons is fixed, the detector and crystal geometry have to be varied in concert to retain a focused condition at the detector for a range of X-ray wavelengths. Bragg law, nλ = 2d sin θ , (1.4.1), and a photo-detector for measuring their intensity. Thus, for a given diffracting crystal with well-defined inter-planar spacing, d, and a given angle of incidence, θ , the wavelength of the unknown characteristic X-ray line can be determined. In addition, because of its finite size, the detecting single crystal is curved (ground or bent) to a given radius, focusing the diverging X-ray beam from the specimen to a fixed spot on the detector. Further, since the take-off angle of X-rays from the specimen surface is fixed, the crystal and detector are moved together to maintain focus, as the angle of the detector is changed to cover a range of wavelengths (Fig. 2.5.1b). Normally, the maximum achievable θ in WDS is about 73◦ . From Bragg’s law, the maximum wavelength of X-rays that can be diffracted/analyzed by the crystal is 1.91d. To detect light elements with long wavelengths (low energy), crystals with large inter-planar spacing are required; in fact, synthetic multilayered materials, with artificial spacing, d, such that 25 < d < 60 Å, are prepared specially for detecting light elements (4 < Z < 9). Further, the resolution of WDS can be obtained by differentiating Bragg’s law, i.e. dθ n = dλ 2d cos θ

(2.5.1)

Thus, using crystals with smaller d-spacing gives better resolution. On the other hand, using a small d-spacing, will restrict the range of wavelengths that can be analyzed. Hence, in practice, a crystal with a larger d-spacing is used to scan the specimen to identify characteristic X-ray peaks of interest; then, a crystal with smaller d-spacing is used to obtain a high-resolution spectrum over a narrower range of wavelengths. Note that the geometry of WDS restricts a specific crystal with a well-defined set of lattice planes, or d-spacing, to detect only a small range

OUP CORRECTED – FINAL

X-Rays as Signals: Core-Level Spectroscopy with X-Rays 109 Table 2.5.1 Diffracting crystals used in wavelength dispersive spectrometers and their range of applicability. Crystal

Plane

Spacing, d (Å)

Atomic range (Z) K-lines

Atomic range L-lines

LiF LiF NaCl Pentaerythritol W-Si Multilayer (LSM)

(220) (200) (200) (002)

1.424 2.014 2.82 4.371 variable

> Ti (Z=22) K(19)–Br(35) S(16)–Rb(37) Al(13)–K(19) Be(4)–F(9)

> La (Z = 57) > Cd (Z = 48) > Mo (Z = 42) Be(4)–N(7)

(b)

(a)

Ti-Kα

EDXS

1.74 keV

Counts (a.u.)

Counts (a.u.)

WDS

Ba-Lα

Ti-Kα Ba-Lβ1

Ti-Kβ

(1) 1.74 keV

4.4

4.8 5.2 Energy (keV)

5.6

(3)

Cu-Kβ

(2)

Ti-Kβ Ba-Lβ2 Ba-Lγ1 4.0

Cu-Kα

1.74 keV

6.0

2

3

4

5 6 7 Energy (keV)

8

9

10

Figure 2.5.2 (a) Comparison of wave length and energy dispersive X-ray spectra of BaTiO3 over identical energy ranges. Notice the much better energy resolution in WDS compared to EDXS. The natural line width of an X-ray peak is ∼2 eV. Measured values are 3.0 eV (WDS) and 150 eV (EDXS) for Ti-Kα . (b) Energy dispersive X-ray spectrum of a Cu–Ti alloy, showing the escape peaks from (1) Ti-Kα , (2) Ti-Kβ , and (3) Cu-Kα X-ray emissions. of wavelengths, typically of the order of a few Angstroms, covering only a small range of elements. Therefore, spectrometers are built to accommodate multiple diffracting crystals, which can be easily interchanged to cover the broadest range of wavelengths and elements. Table 2.5.1 is a partial list of crystals used in WDS and the range of elements whose characteristic X-rays that each one can detect. WDS is characterized by high resolution and high signal to background (Fig. 2.5.2). Its principal limitations are its bulkiness, as multiple analyzing crystals need to be available in the same spectrometer to cover the energy range required for a complete analysis. Further, it involves moving parts that are susceptible to breakdown; moreover, the data acquisition is sequential and very slow. It also suffers from the possibility of higher order Bragg peaks overlapping with other

OUP CORRECTED – FINAL

Atomic Structure and Spectra characteristic X-ray lines, complicating the interpretation. An oft-cited example is S Kα (n = 1) at λ = 5.372 Å, overlapping with Co Kα (n = 3) at λ = 1.789 Å. 2.5.1.2

Energy Dispersive X-Ray Spectrometer (EDXS)

EDXS uses a solid-state semiconductor crystal to simultaneously detect all the X-rays emitted from a specimen. In this way, they are able to overcome a significant limitation of WDS, which involves sequential detection as a function of wavelength, and often requires multiple analyzer crystals to cover the broad range necessary for chemical analysis. Energy dispersive detectors have the advantage of being compact, which allows them to be installed even in the small confines available between electromagnetic lenses in a TEM/SEM column (Fig. 2.5.3a). Figure 2.5.3b shows a schematic arrangement for EDXS. It includes a semiconductor detector, processing electronics, and a multichannel analyzer (MCA) for the display. The semiconductor detector can be either a Li-drifted silicon (Si–Li) or a germanium (Ge) crystal. In the case of an Si–Li detector, the Li is ionimplanted to uniformly compensate any intrinsic defects in silicon. To optimize the performance of the semiconductor detector (Fig. 2.5.4), it is often cooled to liquid nitrogen (L-N2 ) temperatures. Since the specimen to be probed by electron

Liquid nitrogen

X-ray or Electron Beam

FET

Preamplifier

Amplifier

X-rays Printer Specimen

Si(Li) detector

Be window

Keyboard

Multichannel analyser

Display

Ni Kα

Intensity

110

Computer V Kα Cr Kα Co Kα Cr Kβ 4.0

6.0

Ni Kβ

8.0

10.0

Energy (keV) Figure 2.5.3 (a) An EDXS detector installed in a TEM column; only the liquid-N2 Dewar (seen) is outside the column. (b) A schematic representation of an EDXS detection system, including signal processing and display.

OUP CORRECTED – FINAL

X-Rays as Signals: Core-Level Spectroscopy with X-Rays 111 X-rays

–500 V

To FET p-type region (dead layer ~100 μm)

Gold contact surface (~20 nm)

Li-drifted intrinsic region

n-type region Electrons Holes

Gold contact surface (~200 nm)

or X-ray incidence is kept in vacuum, the cooled detector often ends up having water vapor or hydrocarbon contamination on its surface. Thus, to protect the detector, it is isolated from its environment by sealing it with a window that allows substantial X-ray intensity to be transmitted through. The standard window is made of a thin foil (∼7 μm) of beryllium (Be), but even this thin Be window filters X-rays of energy < 1 keV (Z∼11, Na Kα). Thus, to detect low atomic number elements, such as B, C, N, and O, of importance in polymer and ceramic materials, an ultrathin window (∼100 nm of polymer +Al, or diamond-like carbon film) that can transmit X-rays down to ∼150 eV (i.e. detect B Kα = 192 eV) or even a windowless detector, provided the “column” has a good enough differential pumping system to prevent contamination, is employed. When X-ray photons interact and transfer energy to a semiconductor crystal, they produce electron–hole (e–h) pairs by transferring electrons from the valence to the conduction bands. For Si–Li crystals at L-N2 temperatures, the energy required to create a single e–h pair is Ee−h ≈ 3.8 eV, which is a statistical average and much larger than the energy of the band gap. Therefore, a single characteristic X-ray photon of energy, E K , produces Ne−h = EK /Ee−h electron– hole pairs; in other words, the number, Ne−h , of e–h pairs is proportional to the characteristic energy of the X-ray photon. To detect these e–h pairs, the Si–Li crystal is sandwiched between two ohmic contacts (Au coatings), producing a p-type layer on one surface and an n-type layer on the other, with an intrinsic region of the Si–Li crystal in between. A reverse bias applied to this p-i-n junction type crystal detector, separates the e–h pairs and a charge pulse, with magnitude proportional to the number of e–h pairs generated by the specific X-ray photon, is registered. The current signal is amplified, converted to a voltage, and then sent to the MCA to display the counts as a function of energy.

Figure 2.5.4 An EDXS detector showing all the principal components. Adapted (1996).

from

Williams

and

Carter

OUP CORRECTED – FINAL

112

Atomic Structure and Spectra

Figure 2.5.5 Efficiency of detection as a function of energy for Si–Li and intrinsic Ge detectors. Adapted (1996).

from

Williams

and

Carter

Fraction absorbed or detected

1.0 0.8 0.6 0.4 Intrinsic Ge Si (Li)

0.2 0.0 0

20

40 60 Energy (keV)

80

100

Si–Li detectors drop off in efficiency at 20 keV, as higher energy X-ray photons are simply transmitted undetected through the thin detector without creating e–h pairs. However, if an intrinsic Ge crystal of much higher atomic number replaces the Si–Li crystal, it can detect X-rays up to ∼80 keV (Pb Kα) with good efficiency (Fig. 2.5.5). In addition, Ee−h = 2.9 eV, for intrinsic Ge and so it produces more e–h pairs for the same X-ray photon. The EDXS signal is also dependent on two parameters of the signal processing electronics: time constant and dead time. The time constant, τ , is the time window (10–50 μs) the processor allots to evaluate the magnitude (Ne−h ) of the signal pulse. If τ is large, the system can better assign an energy to the pulse, improving resolution, but at the expense of processing fewer pulses in a given time. However, if τ is small, it will help with the statistics (more total counts per second), but at the expense of resolution. To discriminate between pulses, the electronics switches off the detection for a short period of time, called the dead time, immediately after a pulse is detected. If the number of photons arriving at the detector is large, then the dead time would be unacceptably high; in that case, it is best to change the incident probe (X-ray or electron) intensity to reduce the total characteristic X-ray emission. Figure 2.5.2a compares WDS and EDXS spectra from the same specimen of BaTiO3 . The theoretical resolution of a Si–Li detector is ∼110 eV for a Mn Kα (5.9 keV) radiation. In practice, the best EDXS detectors have achieved ∼130– 140 eV (Si–Li) and 115 eV (iGe) for the same Mn Kα radiation; this is very poor compared to WDS (∼3 eV). Two important artifacts observed in EDXS are a sum peak (two X-ray photons arriving simultaneously, but detected as one with twice their energy) and in a Si–Li detector, an escape peak at E k − 1.74 keV, which is caused by the incoming X-ray photon first fluorescing a Si Kα X-ray photon (Kα∼1.74 keV) in the detector before being detected with a correspondingly reduced energy (Fig. 2.5.2b).

OUP CORRECTED – FINAL

X-Rays as Signals: Core-Level Spectroscopy with X-Rays 113

2.5.2

Chemical X-Ray Microanalysis

Characteristic X-rays can be produced either by photon or electron incidence on the specimen. However, the underlying approach for quantitative chemical microanalysis of materials, by measuring the emitted characteristic X-ray intensities, is common to both methods. 2.5.2.1

Photon Incidence: X-Ray Fluorescence Spectroscopy (XRF)

XRF is a non-destructive method for chemical analysis of materials. It irradiates a specimen with a high-energy primary X-ray beam that, in turn, fluoresces characteristic X-ray radiation representative of the chemical composition of the specimen. The characteristic X-ray emissions can be analyzed by either EDXS (Fig. 2.5.3) or WDS (Fig. 2.5.1) detection (§2.5.1). A laboratory XRF system uses an X-ray tube of high power (0.5–3.0 kW) and voltage (30–50 kV) as the source to ensure that the primary X-ray beam (the probe) has sufficient energy to excite the characteristic X-ray photons of interest (the signal) from the specimen. Typically, the target metals in the sealed tube are W, Cu, Mo, Cr, Rh, and Ag, and the tube voltage is adjusted accordingly. XRF does not require a vacuum, although it is preferred for best analysis, and can also function in atmospheres of air and helium. In particular, the ability to detect low-Z elements is affected by the atmosphere, e.g. using EDXS detectors, XRF can detect Z > 6 (C) in vacuum, Z > 11 (Na) in helium, and Z > 13 (Al) in air. Specimens for XRF can be liquids, powders, or bulk; the main requirement to obtain an analysis that is truly representative of its composition is a physically flat surface, and good chemical homogeneity, particularly near the surface, as the primary X-ray beam illuminates a large surface area (∼5 cm2 ) and penetrates ∼50 μm into the specimen. Strictly speaking, the penetration depth of the primary X-ray beam can be controlled by the angle of incidence. At total internal reflection (shallow angles of incidence) it is a few nm, but increases to several μm at large angles of incidence. XRF analysis can be quantitative (§2.5.2.3) and at larger angles of incidence can detect concentrations > 0.1%. For trace element analysis, a grazing incidence angle (few tenths of a degree) with detection limit down to 1013 atoms/m2 , called total reflection XRF (TRXF) [2] has been implemented (Fig. 2.5.6a). TRXF is routinely used as a monitoring tool in semiconductor fabrication to satisfy the stringent demands of surface purity (metallic contamination) on semiconductor (typically, silicon and to a lesser extent, gallium arsenide) wafer surface. The grazing angle of incidence is kept below the critical angle, φ c, given by √ φc = 3.72 × 10−11 ne /E

(2.5.2)

for total internal reflection, where ne is the electron density (cm–3 ) in the specimen, and E is the energy (keV) of the incident X-ray photon. For Mo Kα,

OUP CORRECTED – FINAL

Atomic Structure and Spectra Energy Dispersive Detector

(b)

X-ray Tube X-ray Fluorescence Scintillation Counter Optically Flat Sample (x,y,z, tilt stage)

Si Particles

Coating

(a)

Relative fluorescence intensity

114

φc

Totally Reflected Beam 3 1 2 Incidence angle (°)

Figure 2.5.6 (a) A schematic arrangement for total reflection X-ray fluorescence. (b) Analysis of a silicon surface with a thin layer of nickel, both as a uniform coating and dispersed as fine particles, as a function of angle around the critical angle for total internal reflection. Adapted from Leng (2013).

E = 17.48 keV, and a silicon surface, φ c ∼1.8 mrad. The angular variation of fluorescence around the critical angle will also depend on the chemical nature of the impurity (Fig. 2.5.6b) and hence a precise control of the angle of incidence is important for good TRXF analysis. Further details of TRXF can be found in appropriate ASTM standards [3]. 2.5.2.2

Electron Incidence: Electron Probe Microanalysis (EPMA)

An EPMA, originally developed by Castaing [4], is similar to XRF, but instead of X-rays it uses high-energy electrons, with focusing electron-optics, as the probe to generate and measure characteristic X-ray photons with good spatial resolution (∼100 nm–1 μm). In principle, EMPA instruments are similar to standard scanning electron microscopes (SEMs) equipped with EDXS detectors, but they typically generally have one or more WDS detectors as well. Therefore, the underlying physical principles and quantification methods presented here also applies to microanalysis with SEMs (see §10.6.1). While the interaction of focused electron probes with solid materials is discussed in detail later (§5.3), here it is important to point out that the volume (radius and depth; in the first approximation we will assume they are the same) of the specimen producing characteristic X-ray photons is larger than the diameter, d p , of the incident electron probe. The X-rays are typically generated from a pearshaped volume under the beam (Fig. 2.5.7). The depth, Rx (μm), of characteristic X-ray generation is given by

OUP CORRECTED – FINAL

X-Rays as Signals: Core-Level Spectroscopy with X-Rays 115

dp

To X-ray Detector ψ

E0 (20 keV) dp

Rx Rx

Ec (Cu K) Ec (Al K) E=0

Ec (Cu K) Ec (Al K) E=0 ρ = 3 g/cc

Normalized X-ray intensity

E0 (20 keV)

ρ = 10 g/cc

0.25

Cu-5% Al; E = 20 keV

0.2

Cu Al

0.15 0.1 0.05 0 0

0.2

0.4 0.6 0.8 Depth (μm)

1

Figure 2.5.7 Schematic comparison (left) of X-ray production with depth in two specimens of different densities. The critical volumes for Cu K and Al K emissions are also shown, which gives an idea of the depth/spatial resolution. The take-off angle, ψ, determines the path length of the characteristic X-rays in the specimen. The plot (right) shows X-ray intensities with depth, but now including absorption. Clearly Al K, even though it is generated deeper in the specimen, is absorbed more and is attenuated earlier. Adapted from Leng (2013).

Rx =

 0.0276A  1.67 1.67 − E E x 0 Z 0.89 ρ

(2.5.3)

where A is the atomic weight (g/mole), Z is the atomic number, and ρ is the density (g/cc) of the matrix, E 0 is the incident beam energy in keV, and E x is the energy in keV required to excite the characteristic X-ray line of interest. Another important factor that determines the accuracy of microanalysis, particularly for low-Z elements, is the take-off angle, ψ,determined by the position of the detector with respect to the specimen. The absorption path length in the specimen for any X-ray photon generated at a depth, z, is z cscψ, and such absorption can be minimized by making ψ as large as possible, within the constraints of the instrument. Note that, in an EPMA or SEM, the ability to focus and scan the primary beam over the surface, makes it also possible to produce composition maps of the specimen (Fig. 1.4.14). In summary, EPMA is a quantitative method for elemental analysis of materials in a nondestructive manner. The analysis can be typically carried out for micron-sized volumes at the ppm level. The more sophisticated EPMA systems (Fig. 2.5.8) are capable of simultaneous X-ray (WDS and/or EDXS detectors) analysis, SEM (§10), and back-scattered electron (§10.3) imaging.

OUP CORRECTED – FINAL

116

Atomic Structure and Spectra

Ion Pumps

Electron Source Optical source & CCD camera

Diffracting Crystals ( 1-4) WDS

Beam Stabilizer Airlock Specimen

Figure 2.5.8 A commercial EPMA machine (CAMECA, SXFive) and a schematic cutout illustrating its principal components. 2.5.2.3

Quantitative X-Ray Microanalysis

X-ray spectra can provide qualitative analysis of the elements present in the specimen with relative ease. Characteristic X-ray emission lines (K, L, M, . . .) for all elements are tabulated (Table 2.3.1), and using the software for analysis provided by the manufacturer is a trivial exercise to match and label the observed peaks. Even if the peaks overlap, the relative positions and intensities of the α and β lines can help to identify the elements present. However, we are often interested in the composition or the concentration of the different elements present in the specimen and hence methods for quantitative analysis of the X-ray spectrum are required. In general, the concentration, C A , of an element A of interest is related to its observed characteristic X-ray peak intensity, I A , corrected for instrument, X, and specimen or matrix, M, factors, i.e.CA ∝ IA X M. The instrument factors include details of the source (intensity or current), the geometrical arrangement of the detector (takeoff angle, collection angle) and detector characteristics (efficiency of detection, detector dead time). The matrix factors are threefold: 1. Atomic number effect (Z): This effect relates to how the beam changes with depth as it propagates in the specimen. In XRF, the main concern is the primary absorption of the incident X-ray beam before it reaches the atom to be excited. For electrons there is a similar effect that involves the stopping power of the electron in the specimen; in addition, electrons are back-scattered, and those that are do not engage in characteristic Xray production. For example, a significant fraction (∼15% in Al, ∼30% in Cu, and ∼50% in Au) of the incident beam is backscattered without

OUP CORRECTED – FINAL

X-Rays as Signals: Core-Level Spectroscopy with X-Rays 117 generating X-rays. This loss in X-ray generation increases with increasing atomic number of the specimen. 2. Absorption effect (A): As described, X-rays are generated over a range of depths and as they propagate towards the detector, they are absorbed or scattered in the specimen itself. The intensity of the characteristic X-ray, produced at depth, z, leaving the specimen follows (2.4.2), and varies with the distance given byz csc ψ, i.e.

 I μ = exp − ρz csc ψ I0 ρ

(2.5.4)

Figure 2.5.7 also shows a plot of the sampling depth of Cu Kα and Al Kα radiation in a Cu95 Al5 alloy. Even though the Al Kα is produced deeper in the specimen, the self-absorption of the X-rays by the specimen, makes the sampling depth of Cu Kα to be larger than that of Al Kα, because of the difference in their mass absorption coefficients. In short, this effect is also dependent on the specimen composition. 3. Fluorescence effect (F): This arises from the secondary fluorescence of the characteristic X-rays of interest by the characteristic X-rays of other higher atomic number elements, if present in the specimen. Further, higher-energy bremsstrahlung radiation can also contribute to this secondary fluorescence and the overall effect is dependent on the specimen composition. Quantitatively, X-ray intensities, I, can be related to the concentration, C, by comparing a standard, S, of known composition, with the unknown specimen, U, using the relationship IG CU = UG CS IS

(2.5.5)

where G refers to the generated X-ray intensity, corrected for only the instrument factors but not the matrix effects. However, what is measured in an experiment are the actual intensities, IUE and ISE , and the ratio kEUS = IUE /ISE , also known as the k-factor, which would be different from the generated intensities in (2.5.5), due to matrix effects. Interactions of high-energy X-rays or electrons with solid materials is well understood, and analytical expressions for the Z, A, and F correction factors exist. Therefore, a quantitative ZAF correction is incorporated in all X-ray microanalysis software; it is important to re-emphasize that all three corrections depend on the specimen composition and can be calculated if that is known. Including these corrections, and modifying (2.5.5), we can write the composition of the unknown as   IE CU = Z A F UE CS = Z A F kEUS CS (2.5.6) IS

OUP CORRECTED – FINAL

118

Atomic Structure and Spectra Now, to perform a quantitative analysis, an approximate composition of the unknown specimen is assumed. Based on the assumed composition, the ZAF correction factors are calculated, and a first value, k1IS , for the k-factor is predicted. The difference k1US = k1US − kEUS is computed and in this manner the composition is iterated until the difference, knUS , is below an acceptable value to converge on the quantitative analysis. Since the k-value in (2.5.6) is a ratio of intensities for the same X-ray energy, in the unknown and standard, it is not instrument-dependent, nor does it require detailed knowledge of ionization crosssections or fluorescence yields. Note that microanalysis with X-rays becomes much simpler when we can ignore the ZAF corrections, such as in a TEM where we use very thin electron transparent specimens (§9.4.3). In that case, k-factors for the elements are measured or calculated with respect to Si, i.e. kASi . For example, in the case of a binary alloy, Ax By , we know that CA + CB = x + y = 1. Further, it is easy to relate the measured characteristic X-ray intensities (I A , I B ) to the specimen concentrations (C A , C B ), simply as follows: kASi CA IA CA = kAB = IB CB kBSi CB

(2.5.7)

Thus, if the k-factor, kAB , is known or can be experimentally determined, by measuring the characteristic X-ray intensities (I A , I B ), the composition, (C A , C B ), can be determined (all the other factors cancel out). This approach is discussed in Example 2.5.1, and further in §9.4.3. Example 2.5.1: Microanalysis using k-factors. You are trying to determine the composition of an unknown binary alloy specimen, Ax By , in thin film form. You measure the following X-ray characteristic peak intensities: IA = 10, 000 and IB = 12, 000 counts. You also measure the characteristic X-ray intensities of two available standards, A50 Si50 , and get IA = 10, 000 and ISi = 1, 000 counts, and for the other, B0.67 Si0.33 , you measure IB = 60, 000 and ISi = 10, 000 counts. (a) Determine the k-factors, kASi , kBSi , and kAB . (b) What is the composition of the alloy specimen, Ax By ? Solution: We neglect the ZAF corrections as the alloy is in thin film form. Then we apply (2.5.7) to the two standards. For A50 Si50 , IA /ISi = kASi CA /CSi = 10000/1000 = kASi (50/50). Thus, kASi = 10. CB 67 Similarly, for B0.67 Si0.33 , IIBSi = kBSi C = 60000 10000 = kBSi 33 . Thus, kBSi = 3. Si Further, kAB = kASi /kBSi = 10/3 = 3.33 Now IA /IB = kAB CA /CB = 10000/12000 = 3.33 (CA /CB ), which gives CA /CB = 1/4. But CA + CB = 1. Hence, CA = 0.2 = x and CB = 0.8 = y, and the compound is A0.2 B0.8

OUP CORRECTED – FINAL

Surface Analysis: Spectroscopy with Electrons 119 Photons, E = hf Ec

Ec – δE

Escape Depth δE – energy loss

2.6 Surface Analysis: Spectroscopy with Electrons We now introduce two of the principal surface analysis techniques, AES and XPS, which involve the detection of electrons with energy in the range of 5–2000 eV. A more extensive discussion of these two methods can be found in Watts and Wolstenholme (2003) and Briggs and Seah (1990). In this energy range the electrons have a high probability of inelastic scattering, with an associate loss of energy, δE, as their mean free path lengths, in most solids, are less than 100 Å (Fig. 1.3.5). Thus, if we are able to detect electrons of energy, E c , which is known to be unchanged as it emerges from the specimen, we can safely assume that it originates within a shallow layer (called the escape depth) of the specimen surface. In other words, these techniques are surface sensitive or even surface specific (Fig. 2.6.1). Second, as both the incident electrons and X-ray photons used as the probe in AES and XPS, respectively, have reasonably high energies (∼1.5 keV), they can penetrate and interact with the inner shell electrons of the atoms and, as such, are useful for elemental identification, albeit only at the surface of the specimen. Note that if lower-energy electrons or photons are used as the probe they interact with the outermost, less-tightly bound electrons that are associated with the chemical bonding and are not necessarily associated with a specific atom. Third, since these techniques analyze the surface, any surface contamination, including gas molecules, would adversely affect the signal intensities. Hence, these measurements require clean surfaces, and are carried out in a chamber held at ultra-high vacuum, 10–6 –10–8 Pa (10–8 –10–10 mbar). Finally, all such techniques, including XPS and AES, require some form of electron energy analyzer, and an understanding of the basic principles involved in the construction and use of such analyzers (§2.6.1.2) would help significantly in experiment design and improve metrology.

Figure 2.6.1 Energetic photons incident on the surface of a specimen ionize atoms and create characteristic photoelectrons deep inside. Only those photoelectrons within the escape depth of the surface, emerge with their energy, E c , intact. Those from further down suffer an energy loss, δE, and emerge with energy E c – δE.

OUP CORRECTED – FINAL

120

Atomic Structure and Spectra

2.6.1 Instrumentation for Surface Analysis with Electron Spectroscopy 2.6.1.1 Vacuum Chamber and Other Components Figure 2.6.2 shows a schematic of a modern instrument combining both AES and XPS in a single vacuum chamber for comprehensive surface analysis. Figure 2.6.3 shows a commercial XPS system with its principal components. In principle, an SEM column (§10.2) could also be added to image the area of analysis. Moreover, the electron gun used for exciting Auger electrons can also be focused and used to scan the specimen surface. In this manner, maps of the surface Auger electron distribution in two dimensions can be made (Fig. 2.6.9); the technique is then referred to as scanning Auger microscopy. Finally, a sputter ion gun using Argon ions, can be employed either to clean the specimen surface of any contamination and/or to remove very thin layers of the material between the generation of Auger maps. These maps can also be digitally combined to create a three-dimensional Auger electron map of the specimen. Ultrahigh vacuum (UHV, 10–6 –10–8 Pa or 10–8 –10–10 mbar) is required for surface electron spectroscopy to keep the surfaces free of any contamination. It also prevents the low-energy photoelectrons and Auger electrons from being scattered by residual gas molecules in the chamber, thus enhancing their probability of reaching the analyzer. Adsorption of gas molecules on the surface of the specimen is very rapid (a monolayer accumulates in less than 1 second, compared to typical AES/XPS data acquisition time of several minutes) and a major issue of concern for any surface science experiment. To avoid gas contamination the chamber is baked at 250-300◦ C; this helps dislodge most gas molecules attached to the chamber walls to be removed by the pumping system. The X-ray gun, typically for XPS, is similar to an X-ray tube (Fig. 2.4.2) and common sources are Al Kα (1.4866 keV) or Mg Kα (1.2536 keV), with narrow line widths of the order of ∼1eV. The sputter ion gun, with energies in the range 0.5–5.0 keV, in addition to cleaning the surface, can also be used for secondary ion mass spectrometry (see §5.5 for further details on SIMS). The electron gun used for AES is similar to those used in electron microscopy; both lanthanum hexaboride, LaB6 , filaments and field emission sources are now common, and are discussed later (§5.2.2 and §9.2.1). 2.6.1.2

Electrons as Signals: Electron Energy Spectrometers and Analyzers

A common and desirable principle for measuring the number of electrons in a narrow energy window is to design a spectrometer as a band-pass analyzer. This is achieved by creating a dispersing field, which can either be electrostatic or magnetic. When the electrons pass through this field they are deflected, and the degree of deflection would depend on their velocity or kinetic energy. Electrostatic deflection analyzers are more common for surface analysis as they are compact, UHV compatible, do not generate any significant fields outside the analyzer, and

OUP CORRECTED – FINAL

Surface Analysis: Spectroscopy with Electrons 121

Recorder X

Lens voltage

Voltage

+Hemisphere

Voltage

Retarding

Voltage

High voltage

–Hemisphere

Spectrometer Controls

Channel electron multiplier

Hemispherical Sector Analyzer

Electrostatic Transfer Lens

Analyzer entrance and exit slit plate Amplifier

Electron Gun

Ratemeter

Ion Gun

X-ray Source e–

Recorder Y

θ Specimen

C.P.S.

Analysis Chamber (UHV) Electron energy X-Y recorder

Figure 2.6.2 A schematic drawing of a compound AES and XPS system. Adapted from Watts (1990).

OUP CORRECTED – FINAL

122

Atomic Structure and Spectra Electron Analyzer

Camera system for sample navigation Load Lock

Manipulator XYZ with 180° rotation & heating

X-ray Monochromator Ga liquid metal X-ray source XYZ table for source adjustment

Pumping system

Figure 2.6.3 A commercial Omicron-HAXPES system. A schematic of this hard X-ray photoelectron system, with its principal components is shown on the right. are optimal for the low energy (100 keV), such as in an analytical TEM, and are discussed later (§9.4.2.4) in the context of electron energy-loss spectroscopy. The principles governing the design of an electrostatic analyzer can be understood by considering the simple case of two parallel plates set at different potentials (Fig. 2.6.4a), producing an electric field, E x . If electrons of different energy are directed between these plates along the z-direction, they will be deflected by the field with the magnitude of deflection being inversely proportional to their energies. If an aperture or opening is inserted into one of the plates, by adjusting the field as well as the aperture size, electrons of a specific energy can be made to emerge though the aperture and be counted. However, Figure 2.6.4a shows how the degree of deflection of the electrons would also be affected by their angle of entry into the analyzer. A good spectrometer should be capable of focusing electrons of the same energy, but entering the analyzer over a finite range of incident angles, at the same exit aperture. Figure 2.6.4b shows how, in the parallel plate analyzer, this can be achieved by injecting the electrons at a fixed angle. Based on these principles, two designs of analyzers are used in surface science: the concentric hemisphere analyzer (CHA), with a mean deflection of 180◦ (Fig. 2.6.4c), and a cylindrical mirror analyzer (CMA), consisting of two concentric cylinders (Fig. 2.6.4d). Generally, the aperture for the CHA would be such that it accepts a total solid angle of 10–2 steradians; in contrast, the CMA would accept ∼1 steradian (∼100× larger than CHA) giving higher collection

OUP CORRECTED – FINAL

Surface Analysis: Spectroscopy with Electrons 123 (c)

(a) – x

v1 v1

z

v2

y +

(b) v1

–

(d)

v1

Figure 2.6.4 Principle of electrostatic electron analyzers based on trajectories of electrons in a parallel plate capacitor. (a) Electrons introduced parallel to the plates with velocities, v2 > v1 , strike the plate at different positions. However, another electron with velocity, v1 , may arrive at the same point as v2 , if it is injected axially. (b) The same parallel plates, in a plane mirror focusing geometry. Electrons of the same velocity/energy, injected at different angles are brought to the same focal point. (c) A concentric hemispheric analyzer (CHA) with a 180◦ spherical sector, and (d) a concentric mirror analyzer (CMA) with cylindrical symmetry about the axis. Adapted from Woodruff and Delchar (1986).

efficiencies and signal to noise characteristics. Further details of the design of these analyzers are beyond the scope of this book and can be found elsewhere (Woodruff and Delcher, 1986), but here we can summarize their main attributes. A typical CMA with an outer cylinder diameter of 10–15 cm and no retardation, can operate at a resolving power (E0 / E) of ∼200 and a working distance of ∼5 mm; in other words, a CMA can be a high collection efficiency, low-resolution analyzer. A CHA of comparable size can have a resolving power of 1,000–2,000 and a working distance 25–50 mm; thus, it is a low collection efficiency, highresolution analyzer. Figure 2.6.2 shows a CHA as part of the surface analysis system.

2.6.2 Auger Electron Spectroscopy AES is an electrons-in (probe), electrons-out (signal) technique conducted in vacuum. The Auger process (§2.3.3) involves the ionization of core levels by the relatively high-energy (>1.5 keV) incident electrons and the detection of Auger electrons of discrete energy, following the non-radiative decay (Fig. 2.3.6) of the core hole. Auger electrons typically have energies , can be used as the probe, which eliminates near-ultraviolet, visible, and higher wavelengths. In general, photoelectron spectroscopy is implemented using two types of photon sources available in a laboratory: either a soft X-ray source, 1200 < hf < 1400 eV, used specifically for XPS, or a gas discharge lamp (§5.2.1), 10 < hf < 40 eV, typically used for ultraviolet photoelectron spectroscopy (UPS). The escape energies (E pe ) of photoelectrons are 500–1,400 eV in XPS, and ∼17 eV for UPS, if an He I resonance line (hf = 21.2 eV) is used. From Figure 1.3.5, we can infer that for the case of UPS, the energies correspond to the deep drop in the mean free path with energy for inelastic scattering, and hence, the surface sensitivity would also be significantly dependent on the material. The large gap in energy between these two radiations can be filled, if necessary, by synchrotron radiation (§2.4.3), which provides a large band of tunable radiation from the soft ultraviolet to hard X-rays (hf >10 keV); using a suitable monochromator, photoelectron spectroscopy can be performed at any energy within this range. To gain useful information about the binding energies of electrons in the specimen, the photons from the X-ray source should be as monochromatic as possible. Further, a highly conducting metal source, to minimize cooling in the UHV chamber, is preferred for the target. These choices lead to the two materials already mentioned, Al (Kα = 1.4866 keV) or Mg (Kα = 1.2536 keV). Strictly speaking, the Kα line is composed of the Kα 1 (2p3/2 → 1s) and Kα2 (2p1/2 → 1s) lines (Fig. 2.6.11), but even when combined, for Al they are concentrated to give a FWHM ∼1.0 eV. The corresponding FWHM, taking such spin–orbit coupling into consideration for the Kα lines, for other common metal sources are

Figure 2.6.11 The components of the Kα spectrum of Al. Each of the sub-peaks, Kα1 and Kα2, have a FWHM ∼0.7 eV, but they overlap, giving a FWHM ∼1.0 eV for the total Kα peak. Adapted from [7].

OUP CORRECTED – FINAL

130

Atomic Structure and Spectra Al 2s

Figure 2.6.12 (a) An XPS spectrum of a partially oxidized surface of Al with some contamination. Notice that the Delpha photoelectron energy (E PE ) measured and the binding energy (E b ) are simply related as EPE = hf − Eb , neglecting the work function, which is indicated in the plot separately for Al Kα and Mg Kα in the abscissa. (b) The lower binding energy range (0–200 eV) is enlarged. Compare with the binding energies for these elements in Table 2.2.2.

Al Kα

Adapted from Woodruff and Delchar (1986).

Mg Kα 350

Al

No. of electrons

(a)

EB

O ls

Al 2p

(b)

Metal

No. of electrons

Oxide layer with contaminants

Metal Oxide Plasmons

Oxide Plasmons Valence

180 160 140 120 100 80 60 40 Binding energy (eV)

0

Al 2s

Cl 2s

O Auger

20

Al 2p

900 600

800 700 450

700 800 550

600 500

400

300

200

100

900 1000 1100 1200 1300 1400 650

750

850

950 1050 1150

0 EPE (eV) EPE (eV)

Mg ∼0.8 eV, Cr (∼2 eV), Cu (∼2 eV), and Mo (∼6 eV). Thus, Al Kα or Mg Kα emission lines are narrow enough for most analysis and are commonly used. If higher energy sources are used, or a narrower energy resolution is required, a crystal diffraction monochromator (§7.9.5) can be used, but this will generally result in a loss of intensity of the incident probe radiation. Figure 2.6.12a shows a typical photoelectron spectrum for a partially oxidized and partially contaminated Al film, using monochromatic Al Kα irradiation. The spectrum is dominated by a number of sharp peaks associated with the core states of the atoms, from which they emerge without undergoing any further energy loss. Each of the sharp peaks is followed by higher-energy “tails,” following the original photo-excitation of the core holes. Whilst we have argued that the inelastic mean free path of the outgoing photoelectrons defines the surface sensitivity of XPS, the depth of photoionization is dependent on the propagation and absorption of the incident photons. Thus, any resultant loss of energy of the photon, prior to the photoionization event, is reflected in the inelastic tails observed. Figure 2.6.13 shows the variation in binding energy, E b , of the electrons in the atom as a function of atomic number, Z. As expected (2.2.1), the energies increase as Z 2 . The accessible energy levels for Al Kα and Mg Kα are indicated and for Z > 30 only the outer M and N shells can be ionized. On the higher energy side, additional energy loss peaks in Figure 2.6.12b, corresponding to plasmon resonances— collective excitations of the valence band—by the outgoing photoelectron are observed. For further details on plasmons (§9.4.2.2) and inter-band transitions, see Rather (2013).

OUP CORRECTED – FINAL

Surface Analysis: Spectroscopy with Electrons 131

Binding energy (eV)

105

K

104

103

102

N4 N5 N6

10

30

50 70 Atomic number, Z

90

L1 L2 L3 M1 M2 M3 M4 M5 N1 N2 N3

L M N

110

If we look at the photoelectron spectrum on a finer scale (Fig. 2.6.12b) and focus on the Al peaks, in addition to the presence of multiple plasmon losses, we notice a chemical shift in the peaks from the metal to the oxide. When bonding takes place between elements to form a solid (§3.3), electrons are transferred (ionic) or shared (covalent) between the atoms. This changes the chemical environment of a specific element and a spatial redistribution of the valence electron charges around a specific atom. In other words, the potential around a specific atom is slightly modified and their binding energy changes. These shifts in binding energy of core electrons for the silicon 2p line, upon oxidation, and carbon 1s lines in various molecules are shown in Figure 2.6.14a and b, respectively. Example 2.6.2: The C1s binding energies measured in XPS for organic samples shows the following values, depending on the ligand: C–H (285.0 eV), C–N (286.0 eV),C–O (286.5 eV), C–F (287.8 eV), F–C–F (292.0 eV). Can you explain these trends (hint: consider electronegativity values from the periodic table)? Solution: It is reasonable to assume, as a first approximation, that the initial state effects are only responsible for the observed chemical shifts. Thus, as shown in Figure 2.6.14, as the oxidation state of the element increases the binding energy of the photoelectron increases. Similarly, in these compounds, the chemical shift will depend on the electronegativity of the element that binds to carbon. Pauling electronegativity values are H = 2.1, C = 2.5 eV, N = 3.0, O = 3.5, and F = 4.0. In other words, there will be more electron transfer from C, equivalent to increased oxidation, as the electronegativity of the other element increases, which explains the observed trend.

There are other fine structures in an XPS spectrum for which a more involved explanation of the photoelectron emission process is in order. Strictly speaking,

Figure 2.6.13 Binding energy (log scale) as a function of atomic number for all the elements. The region normally relevant for XPS, using laboratory sources, is shaded. Adapted from Feldman and Mayer (1986).

OUP CORRECTED – FINAL

Atomic Structure and Spectra CF4

(a) 103.4

SiO2

99.15

Si-2p

Si

Calculated binding energy (eV)

(b)

Intensity (a.u.)

132

315

CHF3 CO2

HCO2H H2CO

310

CH2F2

CO

HCN C2H2

CH2F C2H4O CH3OH

C2H6 C2H4

305 108

104 100 Binding energy (eV)

96

CH4

290

295 300 Experimental binding energy (eV)

Figure 2.6.14 Examples of chemical shifts observed in XPS. (a) The shift in Si-2p between elemental Si and SiO2 depends on the oxidation state, and is ∼4.25 eV. (b) Experiment and theoretical values of chemical shifts for C-1s in various compounds. the ideal energy, EPE = hf − |Eb |, of the photoelectron is not what is observed. In an atom, when the core hole is created, the other electrons readjust their energy to lower energy states to screen the core hole, and this additional energy, E a , may be transferred to the photoelectron. As a result of this intra-atomic relaxation, the energy of the photoelectron is modified as EPE = hf − |Eb | + Ea . This is what one would expect if the photoionization and photoemission are stable equilibrium processes. In reality, the process is much too fast and the atomic rearrangement is switched on rather suddenly. Then the energy associated with this relaxation may excite a valence electron to an empty higher level—a process known as “shakeup.” The energy for this shake-up may come from the emerging photoelectron, which would then appear as a peak with lower energy. The peaks are also known as “shake-off” if the valence electron is excited to the continuum. Shake-up peaks are particularly strong for some transition metals, rare earths, and aromatic organic systems. In addition to the change in peak position, in elements with valence or core shells containing unpaired electrons, the spin–orbit coupling can produce states with different energies, resulting in additional fine structure, called multiplet splitting, in the photoelectron spectra. Such multiplet splitting is small and would require the use of monochromators to be resolved. Figure 2.6.15 shows examples of shake-up peaks and multiplet splitting, and further details can be found in more advanced texts such as Briggs and Seah (1990) or Watts and Wolstenholme (2003).

OUP CORRECTED – FINAL

Surface Analysis: Spectroscopy with Electrons 133 Example 2.6.3: For the two morphologies of specimens, (a) a surface layer of atom B over atom A, and (b) a homogeneous ordered alloy of atoms A and B, as shown in the following image, what would the intensity ratio IA /IB in XPS/ESCA look like as a function of the X-ray incident angle, θ? (a)

X-rays

(b) X-rays

IA

IA

IB

IA

IB

X-rays

IB

IB X-rays

θ = 0º

IA

θ = 70º

Solution: The XPS or ESCA signal is mainly detected from the surface of the specimen. Thus for (a) we expect IB to drop off and IA /IB to increase rapidly with the angle, θ. However, for (b), a homogeneous specimen, there will be no change in IA /IB with angle, θ. Representative plots illustrating this variation are shown in the following figure. (b)

IA/ IB

IA / IB

(a)

IB θ

θ

OUP CORRECTED – FINAL

Atomic Structure and Spectra

925

Copper (II) Oxide ‘Shake up’ satellites Cu 2p3/2 Cu 2p1/2

935

(b)

Multiplet splitting of Ni 2p3/2

Nickel (II) Oxide ‘Shake-up’ satellites Ni 2p1/2

Intensity (a.u.)

(a)

Intensity (a.u.)

134

945 955 965 Binding energy (eV)

975

855

865 875 Binding energy (eV)

885

Figure 2.6.15 (a) Shake-up peaks observed in copper (II) oxide and (b) multiplet splitting of Ni 2p1/2 observed in nickel (II) oxide. Adapted from Leng (2013).

2.6.4 Surface Compositional Analysis with AES and XPS It is possible to perform quantitative chemical analysis, similar to X-rays (§2.5.2.3) with both AES and XPS. In the case of AES, the yield, dN x , of Auger electrons for the three-step ABC transition (Fig. 2.3.6), at any thickness, dt, and depth, t, for primary electrons with incident energy, E p , can be written as   dN x = IP (t)nx (t)σA (t)RB Ep , t ωx pno−loss (t)dη

(2.6.1)

where I P (t) is the incident electron flux, nx is the number of atoms of the element, x, of interest per unit volume, σA (t) is the ionization cross-section of energy level A, RB (E p ,t) is the back-scattering factor for electrons of energy, E p , ωx is the fluorescence yield or probability of the decay of A to give the specific ABC transition, pno−loss (t) is the probability of no-loss escape from depth, t, d is the acceptance solid angle of the analyzer, and η is the instrument detection efficiency. Thus, the total electron flux at depth, t, is I (t) = IP (t) [1 + RB (t)]. The first two terms in (2.6.1) are the incident flux and the atomic concentration to be determined. The last two terms are dependent on the specific instrumentation. The third term is the ionization cross-section, which can be either theoretically calculated or experimentally measured [7]. Typically, the general trend in electron ionization cross-section as a function of incident electron energy, E p , is for it to rise rapidly above the threshold (given by the core binding energy, E b ) to a peak value at E p ∼3–4 E b , and then falling off slowly at higher energies. Here, the fluorescence yield is the probability that, following an inner shell ionization, an Auger electron

OUP CORRECTED – FINAL

Surface Analysis: Spectroscopy with Electrons 135 1.0

Sensitivity factors

0.5 0.3 0.2 0.1 0.05 KLL

0.03 0.02 0.01

0

10

LMM

20

MNN

30 40 50 Atomic number, Z

60

70

80

Adapted from Flewitt and Wild (2003).

(instead of the production of an X-ray photon) is observed. It is given by ω =1−

1 1 + aZ −4

(2.6.2)

where Z is the atomic number, and a is a constant with large values of a = 1.12 × 106 for K-shells and a = 6.4 × 107 for L-shells. Thus, for low atomic number elements, Auger emission is favored, but as Z becomes large (>35), characteristic X-ray production starts to dominate (Fig. 1.4.5). The back-scatter factor, RB , depends on both electron energy, E p , and the atomic number, Z. In practice, because of this very complicated dependence of Auger yields on a rich variety of factors, quantitative analysis of Auger spectra is carried out using known standards to obtain so-called sensitivity factors (S i ), for any element, i. For a well-defined beam current, the Auger signal from an element is measured and normalized with respect to the signal from a standard, the ratio being the sensitivity factor, S i . For an unknown specimen, an Auger spectrum is measured and after the peaks are identified, appropriate sensitivity factors are applied to give the concentration C i as Ii /Si Ci = Ii /Si

Figure 2.6.16 AES sensitivity factors for quantitative analysis for 5 keV electron probes.

(2.6.3)

The sensitivity factors are published in the literature (McGuire, 1979), and obtained either from instrument manufacturers (Fig. 2.6.16) or measured for the specific instrument. The last method is rather tedious, but it is the most accurate. In any case, compositional analysis using this method assumes that the specimen to be analyzed is spatially homogeneous, which is seldom the case in practice. Quantitative analysis of XPS data, a technique also referred to as electron spectroscopy for chemical analysis (ESCA), follows along the same lines. Similar

OUP CORRECTED – FINAL

Atomic Structure and Spectra Photoelectron cross-section (10–24 cm2)

136

106

106 (a)

2p

106 (b)

3d

2p

3d

4f 105

4d

3p

1s

1s

105

4f

2s

105 4p

3s 4s 104

104

104 4p 3d

5s

4d 5p

4f

5d

103

103 0

10

20

30

40

50

Atomic number, Z

60

70

80

0

10

20

30

40

50

60

70

80

103

Atomic number, Z

Figure 2.6.17 Photoelectron cross-section calculated for different sub-shells for all the elements: (a) the shells most used in XPS, and (b) the complete set of subshells. Adapted from [8].

to AES, the yield, dN i , for a photoelectron peak, at any thickness, dt, and depth, t, for primary X-rays with incident energy, E p , can be written as dN i = Ix (t)ni (t)σi (t)pno−loss (t)dη

(2.6.4)

where I x (t) is the X-ray flux, ni (t) is the number of atoms of element, i, per unit volume, σi (t) is the photoelectric cross-section of element, i, pno−loss (t) is the probability of no-loss escape of photoelectrons from depth, t, which depends on escape depth, λ, and the analyzer angle, d is the acceptance solid angle of the analyzer, and η is the instrument detection efficiency. Note that, for all practical purposes, the X-ray flux is not attenuated over the escape depth of the photoelectrons. Even though the photoelectric crosssections have been calculated for different subshells (Fig. 2.6.17), the absolute quantification of XPS data is complicated. Here again, a simpler approach is to use pure standards and determine the sensitivity factor, and the concentrations, C i , as discussed earlier for AES.

2.6.5

Comparison of AES and XPS

AES and XPS are two of the most widely used surface analysis techniques, as they both have relatively uniform sensitivity to all elements other than H and He. It is also relatively much easier to produce a high-intensity electron beam as opposed to intense X-ray sources in the laboratory (synchrotron sources are exceptions) and so AES is faster in data acquisition and more sensitive than XPS

OUP CORRECTED – FINAL

Select Applications 137 (the appropriate cross-sections have similar values). On the other hand, high electron beam intensities may cause more surface damage in AES; in contrast, XPS is relatively benign. However, as electrons can be easily focused using electrooptical lenses, AES (unlike XPS) can provide high spatial resolution, including 2D maps of surface composition. The quantification process to obtain quantitative surface microanalysis in both cases is complex, but relative concentrations can be obtained with both methods with reasonable degree of accuracy (∼5%) using standards and sensitivity factors. Lastly, the analysis of chemical effects has been highly developed for XPS, and this technique is used widely to probe chemical states of surfaces.

2.7 Select Applications 2.7.1 XRF Analysis of Dental and Medical Specimens It is well known that human tissues contain many kinds of minerals and trace elements that act as catalysts or structural components [11]. Quantitative analysis, including determining the spatial distribution and chemical state of these trace elements, is important for the analysis of metabolic processes. Sometimes these foreign elements lead to lesions; therefore, their analysis can impact diagnosis. For these materials, XRF is particularly suited for elemental analysis and imaging their spatial distribution for the following reasons: X-rays are easily transmitted in air, they seldom damage specimens, including biological tissue, and no pretreatment, such as fixing, dehydrating, or coating with an electrical conductor, is required. Moreover, XRF can analyze wet and thermally low-resistive materials. Finally, microanalysis using XRF often requires a capillary focusing optics (Fig. 2.7.1a), instead of optical lenses, since the refractive index of glass for X-rays is almost equal to one. The XRF spectrum of a paraffin-embedded lung biopsy specimen confirms the presence of tungsten in a patient suspected of having tungsten carbide pneumoconiosis (Fig. 2.7.1b). Synchrotron radiation sources have intensities greater than laboratory sources (X-ray tubes) by several orders of magnitude. Moreover, laboratory sources have high background, whereas synchrotrons produce highly monochromatized X-rays with low background. As a result, XRF spectra using synchrotron radiation shows negligible background, with the possibility of detecting trace (∼ppm) quantities (Fig. 2.7.1c,d). In summary, teeth, soft tissues, and pathology specimens are regularly examined by XRF. For example, such studies have shown correlation between Ca content and tooth demineralization, as well as Zn, S, and Pb content in teeth with the levels of environmental pollution. Further details can be found in this excellent review [11] and the references therein.

OUP CORRECTED – FINAL

Atomic Structure and Spectra

incident X-rays

(vacuum)

Si detector W Lα

fluorescence X-rays

(air)

distance ~1mm

Scattered peak

(b)

X-ray guide tube

CCD

Intensity (a.u.)

(a)

W Lβ

Ni Kα Fe Kα Ti Kα

mylar membrane specimen

transmission X-ray

X-Y stage

0

NaI detector

(c) Conventional XRF

Ca (Ar) Kα

Zn Kβ

10 Energy (keV)

Au Lα

4

(Cl)

Fe Kα

6

8

20

Zn Kα

Ca Kα

Ca Kβ

2

15

10 12 14 Energy (keV)

16

18

20

Scattered peak

Intensity (a.u.)

Zn Kα Ca Kα

5

(d) SR-XRF Scattered peak

Intensity (a.u.)

138

0

2

4

Fe Kα Fe Kβ

Cu

6 8 Energy (keV)

Zn Kβ

10

12

Figure 2.7.1 (a) Schematic diagram of a micro-focused X-ray fluorescence system using a capillary tube to focus X-rays. The inner surface of the capillary tube is designed to be a paraboloid of revolution and the total internal reflection from the inner surface guides the X-rays to focus (b) X-ray spectrum of a paraffin-embedded lung tissue specimen (experimental conditions: 50 kV, 1 mA, Rh target, and 600s/point). Comparison of X-ray fluorescence spectra from identical specimens using (c) a conventional laboratory X-ray tube, and (d) monochromatized synchrotron radiation. Adapted from [11].

2.7.2 Environmental Science: Contamination in Ground Water Colloids There is now substantial agreement that mobile colloids in ground water are important for transporting contaminants in subsurface environments [12]. Combining synchrotron radiation and energy dispersive X-ray spectrometry, elemental

OUP CORRECTED – FINAL

Summary 139 Fe

Ti

Fe Zn

Cr

Cu Ni Zn

K Ti

Well I

Intensity (log sacle) a.u.

Zr

Fe

K

Ti Cr Fe Ni Cr Zn Ti Cu

Zn Fe+

Well II

Fe Fe K Ca Ti Pb

0

5

Fe+

10

Control

15

20

X-ray energy (keV)

composition of individual colloids, and the identity of the contaminants associated with these colloids have been determined (Fig. 2.7.2). Such analysis of colloids, including controls, showed that Al, Si, K, Ti, and Fe elements were primarily of natural origin, but Ca, Cr, Ni, Cu, Zn, Pb, and Zr, and especially Zn, with large peaks were observed in contaminated wells. Further details of these analysis can be found in [13].

Summary We began with a brief description of the electronic structure of the atom, including quantum numbers that describe the state of the electron, hybridization, and spin–orbit coupling. We also introduced the dipole selection rules that govern single electron transitions, and put all this together to explain the emission of characteristic X-rays, following inner shell ionization, and the nomenclature used to describe them. Alternatively, de-excitation processes can also be non-radiative, with the emission of Auger electrons. Auger processes involve three atomic energy levels (as reflected in their nomenclature), leave the atom in its final state with two vacancies or holes, and need not follow dipole selection rules. A third consequence of photon incidence is the emission of characteristic photoelectrons, and their

Figure 2.7.2 Synchrotron radiation X-ray fluorescence spectra of suspended colloids collected from two contaminated wells and a control. In all cases, the background intensity is consistently ∼100 counts. Adapted from [13].

OUP CORRECTED – FINAL

140

Atomic Structure and Spectra energy distribution can be simply related to the binding energies of the electrons in the solid. X-rays are generated in the laboratory by the bombardment of a metal target by high-energy electrons, and typically show characteristic peaks superimposed on a continuous background. High-intensity X-rays are also generated in synchrotrons. X-rays are detected by wavelength dispersion using single crystals of well-defined lattice spacing and applying Bragg’s law of diffraction. To cover the broad range of wavelengths required, multiple crystals are employed. Alternatively, they can also be detected by energy dispersion using semiconductor detectors, albeit with poorer energy resolution. However, unlike the bulky single crystal WDS detectors, the semiconductor EDXS detectors are compact, and are often included even in the narrow confines of electron optical columns, such as in TEMs and SEMs. Characteristic X-rays are used for X-ray microanalysis, and are generated in instruments either with photon (XRF spectrometers) or electron (EPMA) incidence. The intensities of characteristic X-ray peaks can be related to the specimen composition, provided appropriate correction for instrument (details of the source, geometrical arrangement of the detector, efficiency of detection, etc.) and matrix/specimen (atomic number—Z, absorption—A, and fluorescence—F) factors are made. Alternatively, using standards with well-known composition and a method of ratios, or k-factors, allows for routine chemical analysis. The complementary AES (typical energies T1 T1 > 0 T=0

CB

Band gap

Fermi level VB

N(E) Density of states

N(E) Density of states

N(E) Density of states

Figure 3.2.3 Density of states (DOS) in (a) metal, showing the highest occupied state called the Fermi level, EF . (b) DOS in the free electron model for a metal, including the Fermi– Dirac distribution showing the occupancies of the levels at 0 K and at higher temperatures, T 1 > 0 and T 2 > T 1 . (c) DOS in a non-metal, showing the filled valence band and the empty conduction band separated by a band gap. In (a) and (c), the shading shows the occupied levels. The filled levels in materials are typically probed in X-ray photoemission spectroscopy (XPS) and the unfilled levels in X-ray absorption spectroscopy (XAS) and inverse photoemission spectroscopy (IPES).

OUP CORRECTED – FINAL

Absorption coefficient (cm–1)

152

Bonding and Spectra of Molecules and Solids (3.2.1). The photoelectron spectra (§3.8) of many simple metals follow this freeelectron relationship; the calculated maximum filled state or Fermi level, E F , for example, Al ∼11.7 eV, Mg ∼7.1 eV, and Na ∼3.2 eV, matches the photoemission electron spectroscopy (PES) measurements (Al ∼10.7 eV [1], Mg ∼6.1 eV [2], Na ∼2.5 eV) reasonably well. At higher temperatures, the thermal energy excites electrons to higher energy levels beyond E F , and the Fermi–Dirac distribution gives the fractional occupation of levels as:

104



1+e

102

10

1 0.1

1

f (E) =

103

0.3 0.5 0.7 Photon energy (eV)

Figure 3.2.4 The band gaps in semiconductors are typically measured by optical absorption spectroscopy. Such transitions from the valence band to unoccupied states in the conduction band, measured in Indium antimonide, InSb, a semiconductor with a band gap of ∼0.23 eV at 0 K is shown. See also Figure 3.7.1. Adapted from Kittel (1986).

E−EF



(3.2.2)

kB T

Figure 3.2.3(b) also shows this function. In fact, the electron distribution around the Fermi level, as a function of temperature, measured by ultraviolet photoemission spectroscopy (UPS) is in very good agreement with the Fermi–Dirac function (Fig. 3.8.8). In its ground state, a solid insulator or semiconductor will display a filled band known as the valence band, with a band gap, Eg , separating it from the empty conduction band. Simple optical absorption measurements (Fig. 3.2.4) can be used to determine the band gap, which vary from 0.1 eV in semiconductors to ∼12 eV in some ionic insulators (Fig. 3.3.2). A metal, on the other hand, shows a partially filled conduction band and since there is no gap, their optical properties are quite different; in fact, because of the strong interaction of optical radiation with the free electrons in the solid, metals show high reflectivity. Moreover, metallic materials conduct down to low temperatures, but the conductivity decreases with increasing temperature; in contrast, non-metallic materials are nonconducting at low temperatures with some small increase in conductivity as the temperature is increased. Generally, the electrical conductivity is understood in terms of the filling of energy levels in the bands (Fig. 3.2.3). This rudimentary summary of the electronic structure of materials is sufficient to understand related methods of characterization but, for those interested, further details can be found with a physics flavor in Kittel (1986) and Pettifor (1995), or with a chemistry flavor in Cox (1987) and Borg and Dines (1992). This chapter presents throughout a variety of spectroscopy methods (Fig. 3.7.1 offers a summary) used to measure the electronic structure of solids and surfaces, including the occupied and unoccupied levels. However, before we discuss these methods, we need to briefly review the principal types of bonding and how they affect the electronic structure of materials.

3.3 Interatomic Bonding in Solids The primary bonding in solids (Fig. 3.3.1) is classified into three categories: ionic, covalent, and metallic. For each type, the bonding necessarily involves the sharing

OUP CORRECTED – FINAL

Interatomic Bonding in Solids 153 (a)

(b)

Cl– +

Na

Cl–

+

Na

Cl– +

Na

Cl–

(c)

– Si

+

Na

Cl–

Si

–

–

Si

– –

ER

–

Na+ Na+ Na+

–

–

–

Na+ Na+ Na+

Si Si

(d) E

–

–

–

–

–

–

–

Na+ Na+ Na+

R E0 EA R0

Figure 3.3.1 Schematic representation of (a) an ionic bond, such as in NaCl, with local charge transfer from the cation to the anion; (b) a covalent bond, such as in an elemental semiconductor, Si, with charge sharing to stabilize the bond; and (c) a metallic bond, such as in Na, with a “sea” of delocalized electrons holding the positively charged ion cores together in the solid. (d) The general shape of the bond energy curve as a function of interatomic separation, including the attractive, EA , and repulsive, ER , energy components that stabilize the bond. This curve, generally applicable to all bond types, shows the equilibrium bond energy, E 0 , and bond length, R0 . See also Figure 3.4.1. or transfer of the outermost electrons, and specifically, the bonding determines the final electronic structure of the solid. In addition, weaker secondary bonds resulting from dipolar interactions, without any change in the electronic structure of the atoms or molecules that form the solid, are also observed. In practice, such secondary bonds contribute to the stability of polymeric solids (Fig. 3.1.1); however, here we introduce only the three primary bonds, and further details on bonding can be found in Pettifor (1995).

3.3.1

Ionic Bonding

The ionic bond (Fig. 3.3.1a) is readily found in compounds made of metallic and nonmetallic elements, where valence electrons are readily transferred from the metal to the nonmetal. In this manner, all the atoms acquire a stable inert gas configuration, forming either positively (cations) or negatively (anions) charged ions. The energy of interaction between the ions is due to isotropic Coulombic forces, with an attractive component, EA ∝ −A/R, and a repulsive component, ER ∝ B/Rn (Fig. 3.3.1d). Here, A is a constant that depends on the charge of the cation and anion, B is an empirical constant, R is the inter-ionic distance, and the exponent, n, is a constant in the range of 6–9 that reflects the short range of the repulsive interaction. The 1/R dependence in the attractive potential is long range and hence all interactions in the solid, in addition to the nearest neighbor one, must be considered. Thus, depending on the specific crystallographic arrangements in the solid, the summation of all Coulombic interactions, both attractive (typically next-neighbor interactions of opposite charges) and repulsive (next next-neighbor interactions of like charges), results in the constant A being

OUP CORRECTED – FINAL

154

Bonding and Spectra of Molecules and Solids

Figure 3.3.2 The valence and conduction band energies of NaCl can be derived by starting with (a) free ions, then including (b) the Madelung potential, followed by correcting for (c) the electrostatic polarization arising from adding/removing an electron, and finally (d) the inclusion of the bandwidth from orbital overlap. The band gap is ∼7 eV, the width of the valence band (Cl 3p) is ∼2 eV, and the conduction band (Na 3s) is ∼6 eV. Different spectroscopy measurements discussed in this chapter allow for a detailed evaluation of both occupied and unoccupied states arising in materials as a result of their bonding.

–5

Na+ 3s Conduction band

Binding energy (eV)

0

Cl– 3p 5 Na+ 3s

Band gap

10

Valence band Cl– 3p

Adapted from Cox (1986).

15

(a)

(b)

(c)

(d)

replaced by the Madelung2 constant, AM , for the solid. Typically, the lowest energy structures have the largest value of AM , thus favoring arrangements with the largest coordination (consistent with the notion of a non-directional, isotropic ionic bond). Alternatively, from space filling considerations, the coordination number is determined by the radius ratios of the two ions. The model compound, NaCl, illustrates the electronic structure, i.e. the filled and empty levels, of ionic compounds. The fact that each ion has a closed–shell configuration ensures that the associated bands are either empty or completely filled. The filled band (VB) is thus expected to be the Cl 3p orbital with an additional electron, whereas the empty band (CB) is the Na 3s orbital from which the electron has been removed. The final band structure (Fig. 3.3.2d), then involves adjustments for the Madelung potential of the lattice, and the energy arising from the orbital overlaps.

3.3.2

2 Erwin Madelung (1881–1972) was a German theoretical physicist.

Covalent Bonding

Covalent bonding is central to geology (silicates), biology (DNA, proteins, etc.), and the technologically important family of inorganic semiconductors that includes the elements Si and Ge, as well as the compounds GaAs, InSb, and SiC. In solids with covalent bonding (Fig. 3.3.1b), stable electronic configurations are

OUP CORRECTED – FINAL

Interatomic Bonding in Solids 155 achieved by the sharing of electrons. The bonding in the solid is not different from that in small molecules; for example, the bond energy and bond length of C–C bond in diamond is similar to that found in many alkanes. The number of covalent bonds that is possible for a particular atom is given by the 8–N rule, where N is the number of valence electrons. The covalent bond is directional, i.e. it is formed between specific atomic orbitals placed along well-defined directions. Covalent bonding in solids, including the additional constraints imposed by the crystallographic symmetry (§4.1.7), results in the “band” structure. Simply put, the band structure is the distribution of the energy levels in the covalent crystal along different directions of the reciprocal lattice (which is simply related to the real lattice; see §4.2 for details). Measurements of the band structure of semiconductors, which typically show a filled valence band separated from an empty conduction band by the band gap, are a topic of both academic and technological interest. Figure 3.3.3 shows a detailed measurement of the band structure of GaAs, involving the techniques of photoemission and inverse photoemission spectroscopy, as well as angle-resolved PES measurements. Note the direct band gap of ∼1.4eV in GaAs is observed in the measurements. Photoemission spectroscopy to probe the occupied levels was introduced in §2.6 and further details of this technique, and its complement, or inverse photoemission spectroscopy, to probe unoccupied energy levels, will be discussed in §3.8. In practice, many materials have mixed bonding character. The concept of electronegativity, defined as a measure of the ability of the atom to attract electrons, helps quantifying the fraction of ionic character in an A-B bond as   −1 2 % of ionic character = 100 1 − e 4 (xA −xB )

(3.3.1)

where xA and xB are the Pauling3 electronegativities (see Periodic Table of the elements, page 848 for values) of elements A and B, respectively. Example 3.3.1: Calculate the percent ionic character in (a) Na–Cl, (b) Al–N, (c) Mg–O, (d) Si–Ge, and (e) Cu–Zn. Use Pauling electronegativity values. Solution: We look up a table of Pauling electronegativities for the elements and applying (3.3.1) we solve in the form of a table.

xA xB % Ionic character

Na–Cl 0.9 3.0 66.8%

Al–N 1.5 3.0 43.02%

Mg–O 1.2 3.5 73.35%

Si–Ge 1.8 1.8 0%

Cu–Zn 1.9 1.6 2.2%

3 Linus Pauling (1901–1994) was an American chemist who received the Nobel prize in Chemistry in 1954 for “his research into the nature of the chemical bond and its application to the elucidation of the structure of complex substances.” He was also awarded the Nobel Peace prize in 1962!

OUP CORRECTED – FINAL

156

Bonding and Spectra of Molecules and Solids (c)

(a)

(b)

(d)

L







X K





Figure 3.3.3 (a) The band structure of GaAs derived from both photoemission spectroscopy (PES) and inverse photoemission spectroscopy (IPES), above the Fermi level (labeled as 15), obtained by angle-resolved measurements (not discussed in this book). The band structure is simply the distribution of the energy levels in the crystal along different directions (indicated by the symbols L, , , , X, K, and ) of the reciprocal lattice. The solid lines are the result of calculations. (b) Calculated density of states. (c) Measured XPS data, and (d) comparison of the valence region, between experimental data, after background subtraction, and calculated density of states. Adapted from Hüfner (2003).

3.3.3

Metallic Bonding

A relatively simple model of metals involves delocalized valence electrons that are shared by many atoms. In general, metallic solids have a relatively larger number of orbitals than the available number of electrons, making the sharing of the latter energetically favorable. Thus, in metals, a “sea of electrons” holds together the positively charged ion cores distributed periodically throughout the solid (Fig. 3.3.1c). Furthermore, the electrons shield the positively charged ion cores from mutual electrostatic repulsion, further lowering the energy. A detailed description of metals requires an understanding of band theory, and can be found in standard condensed matter texts such as Kittel (1986) and Sutton (1993).

OUP CORRECTED – FINAL

Interatomic Bonding in Solids 157 Figure 3.3.4a shows a calculated density of states for the three elemental ferromagnetic metals, Fe, Co, and Ni. Notice that the density of states is spinsplit, i.e. it is different for spin-up and spin-down electrons, because the exchange interaction—fundamental to the origin of ferromagnetism—-causes a difference in the population of spin-up and spin-down bands for these three elemental ferromagnets. For both Co and Ni, the spin-up band is completely full but it is not so for iron. Hence, even though iron has a larger magnetic moment, it is considered a weak ferromagnet. The degree of exchange splitting, δEx , is proportional (∼1 eV/μB ) to the magnetic moment [3], and can be measured by spin-polarized photoemission (Fig. 3.3.4b), which gives a value δEx ∼ 2.2 eV for Fe. We mention this here only for completeness—further details of such measurements can be found in some excellent reviews [4,5] and in Krishnan (2016). We revisit photoemission and related spectroscopic methods in §3.8 and §3.9, but first, we discuss the vibration modes of molecules, and the related absorption and inelastic scattering methods used for their analysis.

(a) Febcc

0

Density of states, N(E)

–2 2

Cohcp

0 –2 2

Nifcc

0 –2 –5 0 Energy (eV) relative to EF

(b) Spin-resolved photoemission intensities

2

25 10 8

Fe(100) 25

6 4 2 0

δEex 4 2 6 0 Energy below EF(eV)

Figure 3.3.4 (a) Calculated spinsplit density of states for metallic Fe, Co, and Ni. (b) Ferromagnetic exchange splitting, between majority and minority bands of iron, measured by spin-polarized photoemission (not discussed here). Further details of such measurements can be found in Krishnan (2016). (a) Adapted from Papaconstantopoulos (1980). (b) Adapted from Krishnan (2016).

OUP CORRECTED – FINAL

158

Bonding and Spectra of Molecules and Solids

3.4 Molecular Spectra The next three sections introduce different optical spectroscopic methods to identify different molecular species by probing their electronic, vibrational, and rotational characteristics. This is followed by methods to probe the occupied and unoccupied states of solids (§3.7–§3.9). Much of the latter falls broadly in the realm of surface analysis (Watts, 1990) and such measurements also require ultrahigh vacuum.

3.4.1

Vibrational and Rotational Modes

Figure 3.3.1d shows how materials are composed of atoms or molecules bonded together with an equilibrium bond energy, E0 , and bond length, R0 , arising from a balance between attractive and repulsive forces. However, molecules vibrate about their equilibrium position, R0 , and such vibrations of a diatomic molecule can be understood by considering its energy-displacement curve. If the two atoms are separated by a distance, ± R, slightly larger or smaller than R0 , the energy is increased to E + E, and this, in effect, acts as a restoring force. We can fit a parabola to the actual E(R) curve (Fig. 3.4.1a), which serves as a good approximation for small vibrations, ± R, about R0 . The restoring force, FR , of this harmonic oscillator, in the classical sense, is then  FR = −

d2E dR2

 (R − R0 ) = −A (R − R0 )

(3.4.1)

where, the force constant, A, is a measure of the bond strength. The energy, Evib , of this harmonic oscillator, classically, is given by Evib =

1 A(R − R0 )2 2

(a)

(b)

J 4

E Figure 3.4.1 (a) Vibrational energy levels of a diatomic molecule. The depth of the well is E0 = |Ed | + hfvib /2. The number of vibrational levels increases for heavier molecules. (b) Rotational energy levels for a diatomic molecule with moment of inertia, B. Adapted from Sproul and Philips (1980).

(3.4.2)

Parabolic approximation to E(R)

J ( J +1) 20 h2

8 8π2 B

n

0

R 6 4 2 0

Ed

R0

3

12 h2

Actual E(R) curve

6 8π2 B 2

6

h2 8π2 B 2 2 8πh2 B

4

1 0

2 0

E

OUP CORRECTED – FINAL

Molecular Spectra 159 and its classical vibrational frequency, fvib , is

fvib

1 = 2πμ1/2



d2E dR2

1/2



(3.4.3) R=R0

M2 , the reduced mass of the two-particle vibrating system, simplifies where μ = MM11+M 2 the problem of two particles (mass, M 1 and M 2 ) vibrating about a common center of mass, to a single particle of mass, μ, vibrating about a fixed point. In quantum mechanics, this vibration energy of a diatomic molecule is quantized, such that

Evib,n = (n + 1/2) hfvib

(3.4.4)

where n = 0, 1, 2, 3, . . .. All Evib,n are measured in relation to E = 0, when the system is at rest at R = R0 , and fvib is the classical oscillation frequency given by (3.4.3). Thus, even at 0 K, when n = 0, the atoms are vibrating with energy, Evib = 1/2 hfvib , the so called zero-point energy. Example 3.4.1: Consider the linear molecule H2 , with fvib = 1.28 × 1014 Hz. (a) Calculate its zero-point energy. (b) Compare qualitatively the vibration energy levels of H2 , HD, and D2 , where D is deuterium, the heavy isotope 1 H2 of hydrogen. (c) What is the difference in the dissociation energy of H2 and D2 ? Solution: (a) The zero-point energy of the H2 molecule is Evib = 1/2 hfvib = 4.24 × 10−20 J = 0.265 eV. (b) The reduced mass for the molecules is μ(H2 ) = 12 , μ(HD) = 2/3, and μ(D2 ) = 1 From (3.4.3), f vib ∝ 1/(μ)1/2 . Thus, their vibrational energies can be qualitatively related as Evib (H2 ) =  2/31/2 Evib (HD) = 21/2 Evib (D2 ). (c) The difference in dissociation energy is assumed to be the difference in the zero- point energy of the two molecules. Thus, it is given by

 1 − 1/21/2 Evib (H2 ) = 0.293 × 0.265 eV = 0.078 eV.

In fact, structural details of proteins are determined by carefully substituting deuterium for hydrogen in specific locations of the molecule and monitoring the change in vibrational frequency of that specific bond (see §3.10.1).

OUP CORRECTED – FINAL

160

Bonding and Spectra of Molecules and Solids

4 5 4 3 2 1 0

3 2 1 0 J

6 5 4 Evib

4 Erot

3

3

2

2 1 0 J

1 Evib 0 n Electronic

E 2.0 – 10.0 eV s 2x104 – 105cm–1 120 – 600 nm λ Vis–UV

Electronic + vibration

10–2 – 2.0 eV 102 – 5x103 cm–1 1– 100 μm Infrared

Electronic + vibration + rotation 10–5 – 10–3 eV 0.1– 10 cm–1 10 cm– 1 mm Microwave

Typically Eel >> Evib >> Erot

Figure 3.4.2 Molecular energy levels comprised of electronic, vibrational, and rotational contributions. The rotation levels are only shown for vibrational n = 2 states. The transitions between electronic, vibrational, and rotational levels are observed in Vis-UV, infrared, and microwave regions, respectively, of the electromagnetic spectrum.

Typically, for various vibrating molecules, we observe that fvib = 6 × 1012 − 1.2 × 1014 Hz, a frequency in the mid-infrared. In solids, one also encounters lattice vibrations or phonons—a synchronized movement of all atoms in a crystal, typically at lower frequencies ( fvib = 6 × 1011 − 9 × 1012 Hz) compared to molecular vibrations. However, lattice vibrations when compared to molecular vibrations, are more sensitive to changes in temperature. In addition to vibrations, the kinetic energy of molecules is also affected by rotations about their common center of mass. In quantum mechanics, by solving the Schrodinger equation for such rotations of

OUP CORRECTED – FINAL

Molecular Spectra 161 diatomic molecules, we get quantized energy levels, in terms of J , a rotational quantum number: Erot, J =

h2 J (J + 1) 8π 2 B

(3.4.5)

M2 R2 = μR02 . where the moment of inertia, B, is a constant given by B = MM11+M 2 0 The first few rotational energy levels are plotted in Figure 3.4.1b. The total energy of a molecule is a combination of the electronic, vibrational and rotational contributions. To give a sense of scale, energy levels are separated by 2–10 eV for electronic energies, 10–2 – 2.0 eV for vibrational energies, and 10–5 –10–3 eV for rotational energies. Figure 3.4.2 schematically illustrates these energies for a diatomic molecule, with the rotational contribution exaggerated for clarity. The number of vibration modes in a molecule is related to its number of degrees of freedom. Thus, a nonlinear molecule with N atoms, has 3N degrees of freedom and 3N−6 vibration modes (a linear molecule would have one less, i.e. 3N–5 vibrations modes). In complicated molecules, with atoms bonded to more than one atom, and including nonlinear chains of atoms, the vibration modes can be more complex (Fig. 3.4.3). Vibrations along the bond direction are called stretching modes, while those vibrations perpendicular to the bond direction are known as bending modes. These vibration and related modes can be probed by the absorption of infrared (IR) radiation or by inelastic scattering of light, known as Raman scattering, which we discuss in the next section. Note that IR and Raman spectroscopy are normally discussed in terms of wavenumbers (s = 1/λ) in units of cm–1 . Figure 3.4.4 shows two examples of vibrations modes in H2 O and CO2 .

Symmetrical stretching

Asymmetrical stretching +

In plane deformation or Scissoring

+

Out of plane deformation or Wagging

Rocking

+

–

Out of plane deformation or Twisting

Figure 3.4.3 Possible molecular vibration modes for two identical atoms bonded to a different, third atom. Adapted from Flewitt and Wild (2003).

OUP CORRECTED – FINAL

162

Bonding and Spectra of Molecules and Solids H2O

CO2

Symmetric Stretch

s1= 1388 cm–1

s1= 3652 cm–1

Asymmetric Stretch

s3= 2349 cm–1

s3= 3756 cm–1 Figure 3.4.4 Two examples of different vibrations modes, observed in H2 O and CO2 . The associated wavenumber for each mode is also indicated. See also Figure 3.4.12.

Bending

s2= 667 cm–1

s2= 1595 cm–1

Example 3.4.2: The molecule NaH undergoes a rotational transition from J = 0 to J = 1, when it absorbs a photon of 1.2 × 10–3 eV energy. What is the equilibrium separation between Na and H in the molecule? Solution: From (3.4.5) we can see that for the J = 0 to J = 1 transition 2 2 2 the energy E = 8πh2 B (2 − 0) = 4πh2 B = 2h 2 = 1.2 × 10−3 eV = 1.2 × 4π μR0

10−3 ×1.6021×10−19 = 1.9480×10−22 J. The reduced mass of the molecule mH −27 kg. Thus, = (22.989)(1.0078) is μ = mmNaNa+m 22.989+1.078 = 0.9655 a.m.u = 1.55 × 10 H the equilibrium separation,  R0 =

h2 2 4π μE

 12

=

12 2 6.6257 × 10−34   4π 2 1.55 × 10−27 1.948 × 10−22 

= 1.9 × 10−10 m = 1.9 Å.

3.4.2

Ultraviolet and Visible Spectroscopy (UV-Vis)

The absorption of light in the visible (λ ∼ 380 – 700 nm) and ultraviolet (UV, λ ∼ 180 – 380 nm) regions of the electromagnetic spectrum (Fig. 1.3.2) results from excitations of certain electronic states (Fig. 3.4.2) and forms the basis of UV-Vis spectroscopy, which measures how strongly a specimen absorbs different wavelengths of light. In practice, a UV-Vis spectrophotometer sends monochromatic photons of UV or visible light through a specimen, measures the light transmitted (not absorbed) through it, compares it to a reference cuvette,

OUP CORRECTED – FINAL

Molecular Spectra 163 and plots the absorbance as a function of wavelength. Such absorbance may be measured in some specimens of solid phase, but the majority of UV-Vis tests are run in solution. Hence, a typical specimen is dissolved in a solvent and contained in a cuvette made of quartz, glass, or plastic. It is important that the selected solvent and cuvette material be also transparent to the wavelength being studied. Moreover, the concentration of the solution (specimen) is limited by the Beer– Lambert law, which relates the absorbance, A, of the molecule to its concentration, C, and its molar extinction coefficient, ε, as  A = log

I0 I

 = εCl

(3.4.6)

where I 0 is the intensity measured after the light passes through a reference cuvette, I is the intensity of the light transmitted through the specimen, and l is the path length (standardized to 1 cm for most cuvettes) of the light traveling through the specimen. Here, it is assumed that the solution is sufficiently dilute to avoid any absorbing molecule being in the shadow of another one, such that all molecules present in the cuvette contribute to the absorption signal. Needless to say, the Beer–Lambert law breaks down at higher concentrations. From (3.4.6), for any molecule in solution, we can determine either its concentration (if the molecular extinction coefficient is known) or molecular extinction coefficient (if its concentration is known), all other parameters being measured. UV and visible photons have sufficient energy to excite certain electrons and provide details of the electronic structure of molecules (Fig. 3.4.2). Typically, UV-Vis spectra display one main peak (λmax ) characteristic of the molecular specimen, with the highest absorbance and may also include some smaller peaks. As discussed in Chapter 2, probing transitions between atomic levels require energies in the X-ray region. However, when atoms form molecules, the orbitals hybridize, and create σ and π bonding orbitals, σ ∗ and π ∗ anti-bonding orbitals, n nonbonding orbitals with lone pairs of electrons. Moreover, in molecules the separation between the HOMO and the LUMO is significantly smaller than in the case of isolated atomic transitions, making it possible for it to absorb UV-Vis light. In addition, σ-bonded electrons are tightly bound and require higher energies to excite them, compared to the loosely bound π and n electrons. Thus, only lower energy transitions involving π, σ∗ , π∗ , and n molecular orbitals, i.e. π → π∗ , n → π∗ , and n → σ∗ , contribute to the UV-Vis spectrum (Fig. 3.4.5). In addition, as more atoms are added to the molecule, the separation between the HOMO and LUMO levels decreases (Fig. 3.2.2); thus, more conjugated molecules absorb at lower energies or longer wavelengths. In addition, molecules with heteroatoms, such as oxygen, nitrogen, and sulfur, contribute to molecular resonances, increase the effective conjugation of the molecule, and absorb light of lower energies.4 In practice, simple molecules absorb at λ∼200 nm, and larger dye molecules absorb in the visible region.

4 The parts of the molecule that absorb visible light, and are responsible for its color, are called chromophores.

OUP CORRECTED – FINAL

Bonding and Spectra of Molecules and Solids

(a)

(b) σ*

Antibonding orbitals

π* n π* n σ*

LUMO E = 545 kJ/mol

n

HOMO π

Bonding orbitals

Excited Ground state state ETHENE

BUTADIENE

Ground Excited state state

Filled non-bonding levels σ π* σ σ*

Non-bonding E = 690 kJ/mol level

Vacant anti-bonding levels

π π* π σ*

164

σ

Filled bonding levels

Figure 3.4.5 (a) Ground and excited states for isolated (left) and bonded (right) orbitals, and the magnitude of the HOMO–LUMO gap for two different molecules. (b) Relative energies of the bonding, non-bonding, and antibonding levels in molecules, with transitions included in (blue solid, observed) and outside (red hatched, not observed) the range of UV-Vis spectroscopy. Adapted from Harwood and Claridge (2003), and Duckett and Gilbert (2000), respectively.

Example 3.4.3: Show that the HOMO–LUMO transition in butadiene can be probed by UV-Vis spectroscopy. Solution: From Figure 3.4.5, the HOMO–LUMO energy gap, E = 545 kJ/mol for butadiene. The energy gap per molecule is E = E/N0 = 545,000/6.022 × 1023 J. = 9.05 × 10−19 J The wavelength, λ, of the EM radiation required to probe this transition is λ = h c/E = 6.626 × 10−34 J.s. × 299, 792, 458 m s−1 /9.05 × 10−19 J = 2.19 × 10−7 m = 219 nm. This is clearly in the range of UV-Vis spectroscopy (Fig. 3.4.2).

In addition to concentrations obtained by applying Beer–Lambert law, the location (λ) of the absorption peak in a UV-Vis spectrum gives general information about the structure of a molecule (Fig. 3.4.6). In general, the advantage of UV-Vis is the speed with which it can analyze a specimen and newer models, which forego cuvettes and use fiber-optic probe systems inserted into the specimen solution, are incredibly fast with capabilities of scanning through the entire spectral range in less than three seconds. However, the resolution of the UV-Vis spectrum is too poor to resolve the structure of the molecule or the energies of specific bonds. As

OUP CORRECTED – FINAL

Molecular Spectra 165 (a)

(b) Figure 3.4.6 Typical UV-Vis absorption spectra of (a) propane in hexane solvent, and (b) benzene, naphthalene, and anthracene, superimposed to illustrate the effect of increasing conjugation.

Absorbance

Absorbance

1.8 1.2 0.6 0 200

225 250 275 300 325 Wavelength (nm)

300 400 Wavelength (nm)

500

Adapted from Duckett and Gilbert (2000), and Harwood and Claridge (2003), respectively.

(b)

(a)

Absorbance (a.u.)

Absorbance (a.u.)

CdTe CdSe

115Å 83Å 65Å 51Å 43Å 37Å 32Å

CdS

23Å 19Å 16Å 12Å

300

400 500 600 Wavelength (nm)

700

300

600 750 450 Wavelength (nm)

such, it is not a particularly useful spectroscopic tool for stand-alone diagnostics of molecular composition and structure, but it can be used to characterize optical properties of materials such as bandgaps of semiconductor nanocrystals as a function of their size (Fig. 3.4.7). Example 3.4.4: In a UV-Vis experiment with a path length of 1 cm, we observe concentration and absorbance values as shown in the following table: Absorbance Concentration (μM)

0.14 24

0.24 41.2

0.48 82.3

0.70 120

1.04 178

1.4 257

1.65 321

(a) What is the physical significance of the molar extinction coefficient? (b) From the above data, determine the molar extinction coefficient? What are its units? (c) At what concentration does the Beer–Lambert law fail?

Figure 3.4.7 UV-Vis spectra of quantum dots shown for (a) three different semiconductors of 20-30 Å dia. with dramatic shifts in the absorption peaks, and (b) for CdSe as a function of diameter. Adapted from [17].

OUP CORRECTED – FINAL

Bonding and Spectra of Molecules and Solids

Solution: (a) The Beer–Lambert law (3.4.6) can be rearranged as ε = A/Cl. Thus, the molar extinction coefficient, ε, physically represents the amount of light absorbed per unit concentration. (b) We plot the absorbance data as a function of concentration as shown in the image UV-Vis Absorbance

2 Absorbance (a.u)

166

1.5 1 0.5 0

0

100

200 Concentration (μM)

300

400

From the slope of the linear portion of the data, we determine the molar extinction coefficient to be ε = 5.83 × 10−3 M−1 cm−1 . (c) The Beer–Lambert law fails when the concentration approaches ∼250 μM.

Readers interested in learning more about UV-Vis spectroscopy are referred to Harwood and Claridge (2003), and Duckett and Gilbert (2000).

3.4.3 Classical Model of Rayleigh and Raman Scattering When light of frequency, fI , usually from a laser (§5.2.1.2), is incident (probe) on a specimen, it can be scattered (signal). The scattering can be elastic, with the frequency of the scattered light remaining unchanged (referred to as Rayleigh scattering), or inelastic, resulting in some shifted frequency, fS = fI + fint (referred to as Raman scattering). The frequency, fint , can correspond to some internal vibrational, rotational, or electronic transition, as discussed in the previous section. However, the most important one is the vibrational Raman effect, which we discuss here. Typically, in Raman scattering (Fig. 3.4.8), if the scattered frequency is lower (red shifted), it is referred to as Stokes scattering, and if the frequency is

OUP CORRECTED – FINAL

Molecular Spectra 167 Rayleigh line Stokes scattering

anti-Stokes scattering Rotational Raman Effect

Vibrational Raman Band

Vibrational Raman Band

fvib

fvib

Figure 3.4.8 Schematic illustration of the Raman effect, including vibrational and rotational contributions.

fI + fvib

fI

fI – fvib f

E Figure 3.4.9 Schematic of the induced dipole, μInd , in an atom/molecule on the application of an electric field, E. Here R is the radius of the electron orbit, and l is the displacement of the centroid of the orbiting electron from the nuclear charge, +Ze.

–e

–e R

R +Ze

+Ze l

μInd = α E

higher (blue shifted) it is known as anti-Stokes scattering. Raman scattering can be understood physically using a classical model. When a molecule is subject to an electric field, E, its electrons and nuclei respond by moving in opposite directions (Fig. 3.4.9); thus, the applied field induces a dipole moment, μInd . In the molecule, and for small fields, the relationship between the induced dipole moment and the applied field is linear, i.e. μInd = αE

(3.4.7)

Here, the proportionality constant, α, called the polarizability, is characteristic of the molecule. Note that this assumes that the molecule is highly symmetric and the induced dipole is along the same directions as the applied field. In the general case, less-symmetric molecules, E and μInd , are vector quantities and can be along different directions. In that case, the polarizability would become a second-rank tensor, α ij , relating the two vectors E and μInd . To keep matters simple, we assume a linear relationship for the rest of this discussion. If the electrical field oscillates with time, such as in an electromagnetic radiation (Fig. 1.3.3), then the induced dipole moment will also oscillate with time. Now,

OUP CORRECTED – FINAL

168

Bonding and Spectra of Molecules and Solids the intensity of the scattered light is proportional to the square of the amplitude of the oscillating dipole. Furthermore, any internal motion of the molecule, such as vibration and rotation, can modulate this induced dipole moment, resulting in the appearance of additional frequencies. In a classical description, the polarizability now has a static term, α0 , and a sinusoidal oscillating term with amplitude, α 1 , where α = α0 + α1 cos(2πfint t)

(3.4.8)

Further, for any specific vibrating mode, Vi , we can assume  ∂α  α1 = Vi ∂Vi V =0

(3.4.9)

such that when the polarizability does not change with the vibration, i.e. if ∂α/∂Vi |Vi =0 = 0, the Raman scattering effect is not observed. Basically, this means that the induced dipole moment oscillates at a frequency other than that of the incident wave, fI , because of the oscillating polarizability. Now, for an electromagnetic wave incidence, we can represent E = E0 cos(2πfI t), and rewrite (3.4.7) as μInd = αE = αE0 cos(2πfI t) = [α0 + α1 cos(2πfint t)] E0 cos(2πfI t) = α0 E0 cos(2πfI t) + α1 E0 cos(2πfint t) cos(2πfI t) = α0 E0 cos(2πfI t) + α12E0 [cos 2π ( fI − fint ) t + cos 2π ( fI + fint ) t] (3.4.10) The first term is not shifted in frequency and corresponds to elastic Rayleigh scattering (Fig. 3.4.8) of the wave at the same frequency, fI . Now, specifically for vibrational Raman effect, we can set fvib = fint , and then the lower-frequency term, ( fI − fvib ), is the Stokes scattering and the higher-frequency term, ( fI + fvib ), is the anti-Stokes scattering. Note that in this classical formulism, the intensities of Stokes and anti-Stokes scattering is predicted to be the same, but this is not usually observed and Stokes scattering has a higher intensity (Fig. 3.4.8). Details on why this is the case can be found in more specialized texts, e.g. Cothup, Daly, and Wiberley (1990). Figure 3.4.10 shows the energy level diagram corresponding to this twophoton process; both elastic and inelastic scattering can be understood in terms of the energy transfer from the incident photons to the molecules. When light is incident, certain resonance frequencies are absorbed, raising the molecule to some virtual excited state. If the molecule decays to the original level, a photon at the same frequency is emitted; this is the elastic Rayleigh scattering. The molecule may contain energy levels, corresponding to the vibration modes, at energies higher and lower than the energy level to which it was initially excited. Some of these adjacent energy levels may be unfilled because they may have also been

OUP CORRECTED – FINAL

Molecular Spectra 169

Stokes scattering

fI

fI – fvib

Rayleigh scattering

anti-Stokes scattering Virtual states fI

fI

fI + fvib

fvib

excited by the incident photons. The excited atom may decay into one of these adjacent levels, with frequencies larger (anti-Stokes) or smaller (Stokes) than the Rayleigh line. Before we present practical details of IR absorption spectroscopy and Raman scattering, we briefly discuss the complementarity of these two methods. For readers interested in the physics of Raman scattering, further details can be found in Bernath (1995) and Long (1977). Example 3.4.5: A Raman spectroscopy experiment is conducted with incident light of wavelength, λ = 532 nm, and Stokes scattering corresponding to a wavenumber s = 806 cm−1 , is observed. At what wavelength will this signal photon be observed? Solution: The frequency of the incident light, f I = c/λ = 3 × 108 /532 × 10−9 = 5.6 × 1014 Hz. The frequency of the internal vibration is fvib = c/λvib = c s = 3 × 1010 × 806 = 2.42 × 1013 Hz. Thus, the Stokes scattering will be observed at a frequency fStokes = fI − fvib = 5.6 × 1014 − 2.42 × 1013 = 5.36 × 1014 Hz. This corresponds to a wavelength λStokes = c/fStokes = 3 × 108 /5.36 × 1014 m = 560 nm, for the observed photon.

3.4.4 Selection Criteria for Infrared and Raman Activity IR spectroscopy is based on the absorption or transmittance of electromagnetic radiation in the IR region by molecular vibration. If a particular frequency, fIR ,

Figure 3.4.10 Energy level diagram of a molecule showing Rayleigh, Stokes, and anti-Stokes scattering.

OUP CORRECTED – FINAL

170

Bonding and Spectra of Molecules and Solids

Transmittance

4000 cm–1 3000

Adapted from Leng (2013).

1500

1000

500

1500

1000

500

200

CHCI3 Infrared Raman

Intensity

Figure 3.4.11 Complementary infrared transmittance and Raman spectral intensity of CHCl3 as a function of wavenumber (s = 1/λ). Compare the two spectra at 1,210, 750, and 370 cm–1 .

2000

4000 cm–1 3000

CHCI3

2000

0

matches the vibrational frequency, fvib , of a molecule, then the photon can be absorbed and raise the molecule to an excited state; typically, the absorption would be dominated by transition from the n = 0 to n = 1 state in Figure 3.4.1a. Raman spectroscopy, on the other hand, is a two-photon event, involving the absorption of a photon, accompanied by the excitation of the molecule to a virtual state, followed by the decay and emission of a “scattered” photon, modulated by the vibration modes of the molecule (Fig. 3.4.10). In practice, IR absorption and Raman activity are complementary, as shown for CHCl3 molecules in Figure 3.4.11. It turns out that the selection rules for Raman scattering are different from those for IR (and microwave) absorption spectroscopy; some transitions can be observed only through Raman scattering, some only through IR spectroscopy, some through both, and some through neither. There are some broad guidelines to predict if we can expect IR spectroscopy or Raman scattering for a given molecule. For a molecule to be IR active, the specific vibration mode must cause a change in the dipole moment, μ = ql, of the molecule; here, q is the electric charge, positive and negative, at the charge centers of the molecule, and l is their separation. The dipole moment can be intrinsic or induced. Mathematically, IR activity requires that the derivative of the magnitude of the dipole moment, with respect to the relevant vibration parameter, ϑ, proportional to (R—R0 ) in (3.4.1), at equilibrium (ϑ = 0), is nonzero, i.e. 

∂μ ∂ϑ

 = 0

(3.4.11)

ϑ=0

In contrast, for a molecule to be Raman active, the derivative of the polarizability, α, defined in (3.4.7), with respect to the vibration parameter, ϑ, should be nonzero, i.e.

OUP CORRECTED – FINAL

Molecular Spectra 171

Molecule

Mode of vibration

ϑ3 –

+

ϑ1

ϑ2

Polarizability derivative

≠0

=0

=0

Raman activity

YES

No

No

–

ϑ4

Variation of polarizability with normal coordinate (schematic)

Figure 3.4.12 Polarizability and the variation of the dipole moment for CO2 , a linear molecule, in the neighborhood of the equilibrium position for Raman and IR activities. For each of the four possible vibration modes (row 2), a schematic variation of the polarizability (row 3), and dipole moment (row 6), as a function of the spatial coordinate is shown. If the derivative of the polarizability is nonzero (row 4), such as for mode ϑ 1 , Raman activity is expected. On the other hand, if the derivative of the dipole moment is nonzero (row 7), such as for modes ϑ 2 , ϑ 3, and ϑ 4 , IR activity results.

Variation of diploe moment with normal coordinate (schematic) Dipole moment derivative

=0

≠0

Infrared activity

No

YES

≠0

YES

Adapted from Long (1977).



∂α ∂ϑ

 = 0

(3.4.12)

ϑ=0

Figure 3.4.12 illustrates these criteria with the vibration modes of CO2 , which has two stretching modes, ϑ1 (symmetric) and ϑ2 (antisymmetric), as well as two bending modes, ϑ3 and ϑ4 . Figure 3.4.12 shows the derivative of the polarizability and dipole moment, as a function of the vibration parameter, for each mode. Clearly, only the symmetric stretching mode, ϑ1 , where each bond is polar but the vector sum is always zero, is Raman active (the other, ϑ2 , asymmetric stretch, where one bond is compressed and the other one stretched, creates different bond polarities and results in a dipole, is not), and only mode ϑ1 is not IR active (the

OUP CORRECTED – FINAL

172

Bonding and Spectra of Molecules and Solids other three are). In practice, broadly speaking, we can expect in molecules with relatively low symmetry all vibration modes to be both IR and Raman active; however, for high-symmetry molecules, especially with a center of symmetry, the IR and Raman activity are mutually exclusive. Further, vibrations of molecules with ionic bonds, such as OH− , show strong IR activity, but those with covalent bonds, such as C=C, show strong Raman activity. In the next section(s), we discuss practical details of IR and Raman spectroscopy.

3.5 Infrared Spectroscopy All IR spectroscopy measurements use a radiation source with a broad range of frequencies, and after interaction with the molecule, plot some form of intensity as a function of wave number. The IR portion of the electromagnetic spectrum (Fig. 1.3.2) stretches from the long wavelength end of the visible spectrum (λ = 0.7 μm; wavenumber, s = 1/λ = 14000 cm−1 ) to the short wavelength limit of the microwave spectrum (λ = 1000μm, s = 10 cm−1 ). The near IR (λ = 0.70 − 2.5μm, s = 14285 − 4000 cm−1 ) region is not of much interest for vibrational spectroscopy. The mid-IR region (λ = 2.5 − 50μm, s = 4000−200 cm−1 ) is the region where many useful absorptions occur and contains two important regions: (a) in the “group frequency” region, λ = 2.5−8.0μm, s = 4000 − 1250 cm−1 , many of the strongest absorption bands observed can be associated with vibrations of a portion, or functional group, of the molecule, with minimal contributions from the rest of the molecule; (b) in the fingerprint region, λ = 8.0 − 15.4μm, s = 1250 − 650 cm−1 , the common features are associated with single-bond stretching and bending vibrations similar to that shown for CO2 in Figure 3.4.4. Finally, absorption in the far IR region, λ = 50 − 1000μm, s = 200−10 cm−1 , is associated largely with other bending vibrations. In practice, the IR absorption features are measured and then the problem is to determine what molecular species has contributed to the observed spectra. Generally, the practice is to compare the observed spectra to a catalog of recorded spectra for various modes of known molecules; such reference spectra can be found in Cothup et al. (1990), and Table 3.5.1 gives typical values for some common molecules or bonds.

Table 3.5.1 Typical values of stretching and bending vibrations in wave numbers (cm–1 ) Molecule or bond

Stretching mode

Bending mode

C–H N–N H2 O C=O C=C

2,800–3,000 3,300–3,500 3,500–3,800 1,700 1,600

1,600

OUP CORRECTED – FINAL

Infrared Spectroscopy 173

3.5.1 Instrumentation for Raman and IR Spectroscopy There are two kinds of instrumentation—dispersive and interferometry—for IR and Raman spectroscopy measurements. In dispersive systems, the radiation from the IR source is dispersed as a function of wavelength using prisms and gratings. A diffraction grating (§6.6.3), as shown in Figure 3.5.1, is used to disperse light according to its wave number. The incident beam (for the sake of illustration, only two rays, A and A , are shown) is dispersed by the grating in a discrete direction (rays B and B ), based on Bragg’s law, the wavelength of the radiation, and the period of the grating. Typically, the grating is rotated to change the angle of incidence, and in this way, one grating is used to cover a range of approximately 1,000 wave numbers. From the figure it is easy to see that when the path difference between rays B and B satisfies the condition d(sin θi − sin θr ) = mλ

(3.5.1)

where m is an integer, and d is the period of the grating, constructive interference is observed and a particular wavelength is detected (optical gratings are discussed further in §6.6). To define a spectral element, slits are used to select a window of energy by wavelength. The spectrum is acquired sequentially with a measurement time interval to record each spectral element as defined by the dispersive component and slit, to finally give the intensity as a function of wavelength. Table 3.5.2 summarizes the components used in IR spectroscopy, for different ranges of wavelength. In addition to requiring a long time to acquire a complete spectrum, dispersive instrumentation have many moving parts that need synchronized movement with good accuracy (e.g. slit widths must be closely matched and varied with wavelength). To overcome these limitations, a different approach is used (Fourier transform IR (FTIR) spectroscopy), where all spectral information is contained in an interferogram produced by scanning a Michelson interferometer. FTIR has

B A

B

θr

A

θi

d

Figure 3.5.1 Schematic representation of a diffraction grating with periodic grooves separated by a distance, d. See §6.6 for a further discussion on gratings.

OUP CORRECTED – FINAL

174

Bonding and Spectra of Molecules and Solids Table 3.5.2 Components used in IR spectroscopy. Near IR

Mid IR

Far IR

Wavenumber (1/λ), cm–1 Wavelength range, μm IR source (see §5.2.1 for details) Optical setup

12,500–4,000

4,000–200

200–10

0.8–2.5

2.5–50

50–1,000

Tungsten filament lamp

Nernst glower, globar, or nichrome coil

High-pressure mercury arc lamp

1–2 quartz prisms or prism grating double monochromator

Detector

Lead sulfide photoconductor

2–4 diffraction gratings with either a foreprism monochromator or IR filters Thermopile, transistor, or pyroelectric

Double-prism grating up to 700 μm; interferometric spectrometers up to 1000 μm Golay or pyroelectric

Adapted from Willard et al. 1988.

three main advantages—it records all wavelengths at the same time, has much higher signal collection speeds with possibility of repeat scans, and high signal-tonoise ratios—over dispersive methods.

3.5.2 Michelson Interferometer and the Fourier Transform Infrared (FTIR) Method In practice, to work in the mid-IR (finger printing region) of wavelength, λ = 10 μm, which corresponds to a frequency, fIR = c/λ = 3 × 108 /10 × 10−6 = 3 × 1013 Hz, requires specialized detectors and electronic circuits that are fast enough to rapidly sense at the frequencies involved. However, as we later demonstrate, when the signal is modulated using a Michelson interferometer the frequency at which it is to be measured is practically in a more reasonable range (400 Hz). Figure 3.5.2 shows the optical layout of an FTIR setup, with the principal Michelson interferometer component. The incident electromagnetic radiation (note, only the electric component, E, that is of interest, is considered) is given by E(s) = A(s) sin 2π ( fIR t − sx)

(3.5.2)

where, s = 1/λ = fIR /c, and A(s) is the amplitude for wavenumber, s. The intensity of the beam is I (s) = E 2 (s) = A2 (s)sin2 2π ( fIR t − sx)

(3.5.3)

OUP CORRECTED – FINAL

Infrared Spectroscopy 175 Source

I

R

θ

θ

Metal

Mirror, #2

a

Beam Splitter Detector

d2

b

I

n1 n3 θ

Specimen x

Vm

R

I

I

d1

R

I R

c

e n3

I θ

n1 n3 d

Mirror, #1

R

R

I

θi θr

R

f

Figure 3.5.2 The optical layout for a FTIR set up using a Michelson interferometer. The incident beam is split into two by the beam splitter (note, intensity is indicated by the widths of the beam) and the mirrors, with one of them moving at a fixed speed, control the path length of the individual beams impinging on the specimen. The interferogram is then interpreted in terms of its Fourier component. The inset shows various specimen geometries, (a–f ), other than transmission: (a) reflection on a metal surface, (b) multiple reflection, taking advantage of (d) total internal reflection are some of the ways to enhance the FTIR signal. and its average value, over many cycles, is Iavg (s) = A2 (s)/2

(3.5.4)

where the average, < sin2 θ > = 1/2. At the beam splitter, half of the intensity goes into each of the two beams traveling towards the two mirrors. After reflection by the mirror, the two beams encounter the splitter again, and now half of their intensities (i.e. ¼ of the original) is transmitted to the specimen and detector. In other words, the amplitude of each of the two beams going towards the detector is half the original amplitude, i.e., ½ A(s). Let d1 (fixed) and d2 (variable) be the distances of the two mirrors from the beam splitter; in addition, mirror no. 1 is fixed but mirror no. 2 is moved with a velocity, Vm , as shown. At the point, x, where the specimen is placed (as shown) in front of the detector, the electromagnetic radiation for the two beams are E1 (s) =

1 A(s) sin 2π[ fIR t − s (x + 2d1 )] 2

(3.5.5)

OUP CORRECTED – FINAL

176

Bonding and Spectra of Molecules and Solids and E2 (s) =

1 A(s) sin 2π[ fIR t − s (x + 2d2 )] 2

(3.5.6)

The total E(s) = E1 (s) + E2 (s), using the relationship, sin a + sin b = radiation,



a+b a−b 2 sin 2 cos 2 , is E(s) = A(s) sin 2π[ fIR t − sx + s (d1 + d2 )] cos 2πs (d1 − d2 )

(3.5.7)

Setting the path difference, δ, between the two beams to be δ = 2 (d1 − d2 ), the intensity of the combined beam is obtained by averaging the sin2 θ term at a fixed point, x, over a long time as I (δ) =

1 2 A (s)cos2 (πsδ) 2

(3.5.8)

Using cos2 θ = (1 + cos 2θ ) /2, we can rewrite, I (δ) as I (δ) =

1 2 A (s) [1 + cos (2π sδ)] 4

(3.5.9)

Now taking the average intensity, as δ is varied by moving the mirror no. 2, gives Iavg (δ) = A2 (s)/4

(3.5.10)

which is consistent with the initial beam being split into two. If the intensity, I (δ), is recorded as a function of time, as δ is varied the total signal, over all wavelengths, is given by the Fourier cosine transform of the intensity:  Iavg,Total (δ) =

Iavg (s) cos (2π fIR s) ds

(3.5.10)

In other words, by measuring the Iavg,Total (δ), and taking its inverse Fourier transform, we can obtain the intensity, Iavg (s), transmitted at each wavenumber, s. Example 3.5.1: In a typical FTIR instrument the mirror moves with a speed of 0.2 cm/s. Assuming that the wavelength of IR radiation involved is 10 μm, will the frequency to be measured be within practical detector capabilities? Solution: Let us assume that the mirror no. 2 moves with a speed, Vm = 0.2 cm/s. Then for a given time interval, t, the change in path difference is δ = 2Vm t. If the change δ = λIR , we can expect constructive interference. Thus, the frequency, fM , to be measured is fM = 2×0.2×10−2 m 1/t = 2Vm /λIR = 2Vm fIR /c = 2VM s. Thus, fM = 2V λIR = 10×10−6 = 400 Hz, which is well within modern detector capabilities.

OUP CORRECTED – FINAL

Infrared Spectroscopy 177

3.5.3

Practice and Application of FTIR

Figure 3.5.3 shows the implementation of FTIR spectroscopy for a specimen of polystyrene [6] using the Michelson interferometer. First, the interferogram, Iavg,Total (δ), with no specimen (a), and the polystyrene specimen (c), is recorded. Fourier transform (3.5.11) of these interferograms produces the IR spectra for the measurement without (b) and with (d) the specimen. Already, the IR characteristics of polystyrene can be seen in (d). By dividing the spectrum I of polystyrene (d) by the spectrum I 0 of (b), we obtain the transmittance, T, defined as T = I /I0 , as shown in (e). Note that the IR spectrum can also be plotted as the absorbance (A), which is defined as A = −log T.

(a)

(b)

No sample

Signal, I0

0 –3.0 (c)

No sample

(d)

Polystyrene

Polystyrene

3.0 Signal, I

Signal (Volts/scan)

3.0

(3.5.11)

0 –3.0 440 520 Data points

600

(e)

Transmittance

360

Polystyrene 75 50 100 4000

3200 2400 1600 Wavenumbers (cm–1)

800 400

Figure 3.5.3 Interferograms obtained without (a) and with a polystyrene (c) specimen, with the corresponding FTIR spectra, as a function of wave number, s, determined after the inverse Fourier transform, are shown in (b) and (d), respectively. The transmittance spectrum of polystyrene is obtained by dividing spectrum (d) by spectrum (b). Adapted from [6].

OUP CORRECTED – FINAL

Bonding and Spectra of Molecules and Solids The range of frequencies, over which the interferometer can be used is determined by the spacing between data points in (a) and (c). In practice, for a helium–neon laser, this usually works out to 400–4,000 cm–1 , as shown in (b), (d), and (e). Further, the wavenumber resolution, sres , of the IR spectrum, depends inversely on the maximum path length difference, δmax , i.e. sres ∼1/δmax . Thus, a typical FTIR set up, with δmax = 10 cm, gives sres ∼ 0.1 cm−1 , which is more than adequate for most polymeric materials. CH3

OH

=

O H2C = C

C O CH2C CH2O

CH3

H

OH

O =

(a)

O CH2C CH2O C C

C CH3

H

CH2

CH3

(b) .60 (d)

Time (min) 20

.50

46 90

.40 .30 .20 Initial .10 After 20 Min

Absorbance

Absorbrance

178

.125

188

–.125

258

–.175

1155

–.125

.00 –.175 (c)

Difference (×2)

1675 1650 1625 1600 1575 1550 Wavenumbers

1675 1650 1625 1600 1575 1550 Wavenumbers

Figure 3.5.4 FTIR can monitor the curing of polymers used in fabricating polymer-matrix composites, such as bisphenol (a). The FTIR spectrum (b) shows a clear stretching mode of the double-bond at 1640 cm–1 and two modes of ring vibrations, at 1,608 cm–1 and 1,582 cm–1 , to serve as internal standards. After processing, the polymer is loaded and any continued cross-linking of the polymer can be monitored (c) by the reduction in the double-bond stretching mode intensity. (d) This feature, monitored over time, shows that curing continues even for many days after the processing. Adapted from [6].

OUP CORRECTED – FINAL

Raman Spectroscopy 179 Recent developments in fabricating aircraft structures using polymer-matrix composite materials, has made FTIR spectroscopy into an essential industrial tool. It is used to evaluate the post-cure evolution, or the cross-linking of prepolymers used in such fabrication, as a function of the type and amount of initiator used and the molecular flexibility of the pre-polymers. An example of such a polymer is bisphenol (Fig. 3.5.4a), which has a double-bond stretching mode at 1,640 cm–1 , and which will decrease in intensity as cross-linking takes place. In addition, two neighboring bands, at 1,608 cm–1 and 1,582 cm–1 , assigned to benzene ring vibrations, serve as internal standards. The polymers are cured at elevated temperatures, and their initial spectra measured immediately (b). The polymers are then placed in a simulated environment similar to their final use, removed after specific periods of time, and their FTIR spectra measured (b). Additional cross-linking, taking place in the simulated environment even after the processing step, is confirmed by the difference between the two spectra (c). Further, time-dependent evaluation (e) of the polymer cross-linking, shows that the 1,640 cm–1 band continues to increase, suggesting that the curing continues for several days after the initial process.

3.6 Raman Spectroscopy 3.6.1

Raman, Resonant Raman, and Fluorescence

Normally, in Raman spectroscopy (§3.4; Fig. 3.4.4; Fig. 3.4.6), the incident photon is of a wavelength not specifically related to any electronic excitation line (Fig. 3.6.1a) of interest in the material. However, if the incident laser light is tunable, its wavelength or energy can be made to match that of a specific electronic transition (Fig. 3.6.1b) and then the resulting Raman spectrum is further resonance-enhanced with orders of magnitude increase in signal. In addition to signal enhancement, the resonance signal may even be different, with the possibility of the appearance of new bands. An obvious, but important, application of resonance Raman spectroscopy is the ability to work with lower concentrations of the material. Figure 3.6.2 is a good illustration of how the resonance Raman spectrum differs from the normal Raman spectrum for K2 CrO4 . The normal Raman spectrum shows only the four bands characteristic of the CrO4 2– ion; in contrast, the resonance Raman spectrum shows a progression of ten harmonics of the symmetric stretching mode (853 cm–1 ), i.e. increasing bandwidth but with a reduction in intensity. Many colored specimens, with impurities, can absorb the incident laser light and re-emit it as a fluorescence (Fig. 3.6.1c). Unfortunately, the intensity of the fluorescence can be higher by as much as a factor of 104 than the weaker Raman signal, and can completely mask the signal of interest. In fact, fluorescence is considered a major drawback in the application of resonance Raman spectroscopy. Various approaches to overcome this limitation include (a) prolonged irradiation

OUP CORRECTED – FINAL

180

Bonding and Spectra of Molecules and Solids Vibrational energy states

E1

Energy

Virtual State

Figure 3.6.1 Schematic illustration of the Stokes, Rayleigh, and anti-Stokes transitions for (a) Raman and (b) resonant Raman scattering. Often, (c) fluorescence from the specimen dominates and swamps the Raman spectrum.

hf

E0 S R AS (a) Raman

(a)

9

10

S R AS (b) Resonant (c) Fluoresence Raman

8 6785

~7616

6

7

8000

5098

7000

Figure 3.6.2 (a) Resonance Raman spectrum of K2 CrO4 at 363.8 nm excitation, compared with (b) a normal Raman spectrum at 632.8 nm excitation. (c) Absorption spectrum of aqueous K2 CrO4 solution. Adapted from Long (1978).

600 400 200 Wavenumber (cm–1)

4253

3405

3

2 n=1

853 2554 1704

6000 5000 4000 3000 2000 1000 0 Wavenumber (cm–1) (c) 2.0 Absorbance

(b)

4

5942

~8470

9000

5

Vibrational energy states

0

1.6 1.2 0.8 0.4 0.0

30,000 25,000 20,000 15,000 cm–1

of the specimen, prior to the Raman experiment, to bleach out the fluorescence signal; (b) using a larger wavelength laser with lower excitation energy, and (c) using a pulsed laser to take advantage of the shorter lifetime (1–100 ps) of Raman scattering, compared to that of fluorescence (1–100 ns), combined with electronic time-gating to discriminate between the two contributions.

OUP CORRECTED – FINAL

Raman Spectroscopy 181

3.6.2 Instrumentation for Raman Spectroscopy and Imaging A simple laser is used as a source of monochromatic light and a dispersion grating (§6.6.2) arrangement for Raman spectroscopy (Fig. 3.6.3a). In principle, there are two ways to increase the spectral resolution, i.e. the ability to resolve spectral features of such a system. One is by increasing the focal length (Fig. 3.6.3); (a) Specimen z

Reflected Laser Beam

Laser

Scattered Light

y

x

Raman Scattered Light Diffraction Grating

CCD

Filter Scattered Beam (i)

Scattered Beam

Scattered Beam

(ii)

(iii)

Focal length

(b)

Computer

Computer Grating

Camera Optical Microscope

Filter

Specimen Stage y

CCD

x Laser

Figure 3.6.3 (a) Raman scattering using a filter and diffraction grating, for different specimen arrangements, shown in inset include (i) liquid capillary tube, (ii) powder pellets, and (iii) single crystals. The spectral resolution depends on the focal length and the density of gratings. (b) An imaging, micro-Raman arrangement combined with a standard optical microscope. The laser beam can be focused on a specific feature of a specimen, and the Raman backscattered light is filtered to eliminate the Rayleigh line and dispersed with a grating. The x-y specimen stage can be moved to obtain data at points across the specimen. The arrows indicate the beam path from the source (laser) to the detector (CCD) after interaction with the specimen.

OUP CORRECTED – FINAL

182

Bonding and Spectra of Molecules and Solids approximately, doubling the focal length also doubles the spectral resolution. Alternatively, the grating can be changed, and similarly, doubling the density of gratings gives twice the spectral resolution, but higher density gratings have restricted working range; for example, IR radiation is incompatible with gratings of line density larger than ∼2,000 lines/mm. Moreover, as discussed, to minimize fluorescence from swamping out the Raman signal, a longer wavelength laser may be necessary. On the other hand, higher-energy UV lasers might be required to ensure good specimen penetration, especially if fluorescence is not a problem. From a practical point of view, it may be preferable to work with lasers in the visible spectrum. As a result of all these competing factors, a good system will either have a tunable laser, or a number of lasers, that are switchable, along with different filters to eliminate the Rayleigh line, and matched gratings for dispersion and detection. In addition, Figure 3.6.3a shows scattering geometries appropriate for (i) liquids in capillary tubes or glass fibers, (ii) powder pellets or films on substrates, and (iii) oriented single crystals, which gives the maximum information. Direct imaging involves examining the whole specimen for characteristic shifts, e.g. of a single compound. This generates an image showing the chemical distribution of that compound. In contrast, using a focused laser probe, Raman spectra can be taken at points across the specimen, helping identify multiple compounds and their spatial distribution. Generally, such a micro Raman imaging system (Fig. 3.6.3b) is combined with an optical microscope to observe the specimen and identify microstructural features of interest, with connections for the incoming laser light and the outgoing scattered light. Further, a filter is installed in the return optical path after Raman scattering, to remove stray light, specifically the Rayleigh line, and a grating is used to disperse the Raman scattered light into a spectrum. Incidentally, such data acquisition for Raman imaging requires lot of time, computer power, and data storage. Last but not least, Fourier transform Raman spectroscopy, similar to what we have discussed for FTIR, has been developed; details can be found elsewhere (Hendra, Jones, and Warnes, 1991).

3.6.3 Application of Raman Spectroscopy in Chemical and Materials Analysis Table 3.6.1 summarizes the application of Raman spectroscopy to identify common chemical functional groups. Figure 3.6.4a shows a typical application of Raman spectroscopy in materials research. Si1-x Gex is a semiconductor synthesized over a wide range of compositions, and is used for heterojunction bipolar transistors or as a straininducing layer for CMOS transistors. The Raman spectrum from this compound semiconductor shows four characteristic vibrational bands related to Si–Si, Ge–Ge, Si–Ge, and the Si–Sisubstrate modes. Moreover, the position of the Si–Si vibration band, depends linearly on the Ge concentration, x, as shown in

OUP CORRECTED – FINAL

Raman Spectroscopy 183 Table 3.6.1 Raman frequencies of common functional groups. Functional Group

Position (cm–1 )

>S–S< C–C C–C Aromatic ring Aromatic ring CH3 umbrella mode CH3 and CH2 deformations >C=C< >C=C< >C=O mixed with NH deformation >C=O C=C SH >CH2 >CH2 CH3 CH CH CH CH NH OH

500–550 ∼1,060 and 1,127 ∼77–1260 ∼1,000 ∼860 ∼1,375 1,410–1,460 ∼1,650 ∼1,623 1,620–1,690 1710–1745 2,100–2,300 2,540–2,600 2,896 and 2,954 2,845 and 2,880 2,870 and 2,905 2,900 3,015 3,065 3,280–3,340 3,150–3,340 3,000–3,600

Remarks Polyethylene Highly mixed in complex molecule Monosubstituted; 1,3 disubstituted; 1,3,5 trisubstituted 1,4 disubstituted

Ethylene Amide Changes for aldehyde, ketone, and ester

Ethane Polyethylene Polypropylene Cellulose Olefinic CH Aromatic CH Acetylenic CH Broadened and shifted by H bonding Broadened and shifted by H bonding

the calibration plot (Fig. 3.6.4b). Thus, by measuring the line position (510 cm–1 ) of the Si–Si Raman peak in the unknown alloy, ∼120 nm thick, and using the calibration plot, we can determine its composition to be Si0.7 Ge0.3 .

3.6.4 Surface-Enhanced Raman Spectroscopy (SERS) It is well-known that the Raman signal for many molecules, using laser excitation in the visible or near-IR (NIR), is orders of magnitude larger when they are adsorbed on metal surfaces compared to isolated molecules or in solution (see Ferraro, Nakamura, and Brown, 2003). This surface-enhanced Raman spectroscopy (SERS) is a promising development as an in situ vibrational probe for studying liquid–solid, gas–solid, and solid–solid interfaces. Most common substrates used

OUP CORRECTED – FINAL

184

Bonding and Spectra of Molecules and Solids

(a) 350

250

65.5 nm

Si-Si substrate

200 Si-Ge 150

100 nm

520

Si-Si Raman shift (cm–1)

Intensity (a.u.)

300

(b) 121 nm x = 0.3

Ge-Ge

515

84 nm

876 nm 704 nm

510

50 nm 121 nm

505

100 500 200

250

300 350 400 450 Raman shift (cm–1)

500

550

0

0.1

0.2

0.3

x

Figure 3.6.4 Application of Raman spectroscopy in materials research. (a) The Raman spectrum from a 121 nm thick specimen of Si1-x Gex . (b) A calibration plot showing that the position of the Si–Si vibration line varies linearly with the concentration of Ge in the alloy. From the calibration plot, and the observed position of the Si–Si line (∼510 cm–1 ), the composition of the unknown alloy is determined to be x = 0.3. Adapted from Leng (2013).

for SERS studies are colloidal nanoparticles of gold and silver, as well as electrochemically roughened silver electrodes. The fundamental mechanism underlying the SERS effect is still poorly understood. The Raman signal is proportional to the square of the induced dipole moment, μInd = αE. Thus, the enhancement can occur either through changes in the molecular polarizability, α, called the chemical effect, or through the electric field, E, called the electromagnetic effect. The electric field, E, may be significantly changed in the vicinity of a metal surface, particularly when fine particles with significant curvature or rough surfaces are involved. The laser light also collectively excites the valence band—known as plasmon resonance—of the metal, causing the roughness feature of the metal to be polarized and enhancing the local electromagnetic field from that of the applied radiation, leading to Raman signal enhancement. The chemical effect is attributed to a charge transfer or bond formation between the metal and the absorbing molecule, thereby increasing its polarizability. In summary, Figure 3.6.5 compares the relative strengths of Raman scattering, FTIR, and NIR absorption, that can help plan and identify the application of these complementary techniques.

OUP CORRECTED – FINAL

Probing the Electronic Structure of Solids 185 Raman Scattering

Low frequency modes Easy sampling Resonance & surface enhancements possible Narrow linewidths

FTIR Absorption

Good fingerprint libraries available Fundamental vibrations

Water compatible Probe molecular vibrations

Non invasive Fiber optics & remote sampling

NIR Absorption

Figure 3.6.5 Comparison of Raman, FTIR, and NIR methods in materials characterization. Adapted from McCreery (2000).

3.7 Probing the Electronic Structure of Solids Spectroscopy methods routinely provide information about the energy levels in a solid (Fig. 3.7.1). In particular, details of the structure of the valence band and the unoccupied states in the conduction band (in addition to the core levels already discussed; see §2) are of interest. The simplest and commonly used method is that of optical absorption in the UV and visible range, referred to as UV-Vis spectroscopy. Such UV-Vis spectra (Fig. 3.7.1a) of solids shows absorption peaks corresponding to transitions of electrons from filled valence bands to empty states in the conduction band. Specifically, in nonmetallic materials (e.g. semiconductors), this allows for the determination of the band gap (Fig. 3.2.4). However, detailed interpretation of such optical transitions depends on the complex details of both the filled and empty bands, and it is often difficult to resolve them separately. On the other hand, there are other spectroscopic techniques, using X-ray photons or high-energy electrons, that offer opportunities for a straightforward interpretation of the spectrum in terms of the energy levels and their dispersion, i.e. band structure. We explore these techniques in the remaining sections of this chapter. Figure 3.7.1 summarizes the principal spectroscopic methods with photons and electrons, schematically illustrating the techniques and the energy levels involved. Note that we define the vacuum level, EV , as the energy at which an electron is able to escape completely from the electric potential of the solid, and define our energy scale such that EV = 0. However, inside the solid, the normal reference point for energy of the electrons is the Fermi level, EF , already defined in

OUP CORRECTED – FINAL

Bonding and Spectra of Molecules and Solids Kinetic energy

(c)

hf EV, Vacuum level Empty states

0 hf

, work function

hf

EF, Fermi level Filled states Binding energy

186

(a)

(b) hf

hf

Core level (d)

(e)

Figure 3.7.1 This is a key figure that schematically represents the important spectroscopic techniques used to probe the electronic structure of a solid, including (a) optical spectroscopy, see Figure 3.2.4; (b) photoemission, (c) inverse photoemission, see §3.8; (d) X-ray absorption, §3.9, where the core electron can be excited either to the continuum or some empty state, and (e) X-ray emission (discussed in §2). Energy levels of interest, and the work function, are also indicated. Electrons with energies greater than the vacuum level can either leave (b) or enter (c) the solid. Note that in this plot the energy scale is set such that E = 0 at E = E V . Adapted from Cox (1987).

Figure 3.2.2. The difference between these two energy levels is the work function, , which is simply given by the difference as  = EV − EF

(3.7.1)

Thus, the work function is the energy required to remove an electron from the solid into the vacuum (Fig. 5.2.4), which in a metal is typically ∼2–5 eV. In practice, the work function is important for many surface analysis methods, and depends on various surface effects, such as adsorption of different molecules, and surface crystallography (which is introduced in §4.1.7 and discussed further in §8.3). From a practical point of view, work function is also important for thermionic emissions from solids (§5.2.2). Such thermionic emitters release electrons upon heating and are used as electron sources or guns, e.g. in electron microscopes (see §9.2.1 and §10.2.1). The best thermionic emitting materials have low work functions, and the metallic compound lanthanum hexaboride (LaB6 ) is widely used.

OUP CORRECTED – FINAL

Probing the Electronic Structure of Solids 187 The energy scale is defined such that the electrons bound in the solid have negative values and those that are free, having left the solid with sufficient energy to overcome the work function, are assigned positive values (Fig. 3.7.1). The best way to probe the energy levels and the dispersion of electrons in filled levels of solids, especially in the valence band, is by (b) photoelectron spectroscopy. The complementary, inverse photoelectron spectroscopy, as shown in (c), probes the unoccupied, or empty, states. Information about empty states is also obtained from X-ray absorption spectroscopy (XAS) (d). Here, the fine structure seen at the onset of characteristic absorption edges, known as X-ray absorption near edge structure (XANES), and long-range oscillations, known as extended X-ray absorption fine structure (EXAFS), are commonly investigated (see §3.9.2). X-ray photoelectron spectroscopy (XPS) and XAS are complementary methods (Fig. 3.7.2). In XPS, a bound electron is excited to a free state in vacuum with a well-defined kinetic energy; such transitions, shown for inner-shell excitation Electron binding energy

(a)

0

EF L L L

K

Photoelectron intensity

Free electron K-shell

L L L

Conduction Band

Kinetic energy of photoelectrons

EF I0

L L L

K Specimen IT

X-ray absorption, I0 – IT

(b)

Photon energy in incident X-ray beam

Figure 3.7.2 Complementary nature of (a) X-ray photoelectron (XPS) and (b) X-ray absorption spectroscopies (XAS). Adapted from Siegbahn et al. (1967).

OUP CORRECTED – FINAL

188

Bonding and Spectra of Molecules and Solids (K- and L-shells) as sharp lines, can also probe the valence or conduction (shown here for a metal) bands. Note that a constant energy photon probe is used and the kinetic energy of the photoelectron signal is measured. In XAS (Fig. 3.7.2b), an edge occurs when an X-ray photon is absorbed and a bound electron is excited, following dipole selection rules, to the first available unoccupied state above the Fermi level, displaying some fine structure based on the unoccupied density of states; some electrons are excited beyond the vacuum level, with probability decreasing with increasing energy transfer, and appearing as a gradually decreasing tail in the absorption spectrum. In XAS, the absorption is measured as a function of photon energy. We first revisit photoelectron spectroscopy as a probe of the occupied electronic levels in a solid, followed by an introductory description of absorption methods.

3.8 Photoemission and Inverse Photoemission from Solids Section 2.6 introduced the physical principles of photoemission and some details, including instrumentation, of the related technique of photoelectron spectroscopy using incident X-rays (XPS) or ultra-violet radiation (UPS). In principle, all photoemission experiments are performed as described in Figure 2.6.2. In addition to an X-ray tube, and a UV gas discharge lamp (§5.2.1.1), synchrotron radiation sources (§2.4.3) with higher intensities and tunability of the radiation are also used. The light impinges on a specimen and ejects a photoelectron, which is analyzed both in terms of its kinetic energy, Ekin , and its momentum, p. Thus, in experiments, the kinetic energy of the emitted photoelectrons and angle-resolved measurements that specify the collection geometry (Fig. 3.8.1a), and thereby the direction of their momentum, are measured. In a solid specimen, the kinetic energy and momentum of the photoelectron are simply related to the binding energy, EB , (with E B = 0 at E F ) and the work function, , Ekin = hf − − | EB |

(3.8.1)

and p= 5 Pioneered by Kai Siegbahn (1918– 2007) and colleagues in Sweden. Professor Siegbahn shared the 1981 Nobel prize in Physics and was cited for “his contribution to the development of high-resolution electron spectroscopy.”



2me Ekin

(3.8.2)

with the direction of p being determined by the polar, θ , and azimuth, ϕ, angles (Fig. 3.8.1a) along which the electrons are collected by the spectrometer after they leave the specimen. In this specialized case, where the photoelectrons are collected as a function of direction, the technique is known as angle-resolved photoemission spectroscopy (ARPES).5

OUP CORRECTED – FINAL

Photoemission and Inverse Photoemission from Solids 189

(a)

Photon Source

Ekin

(b)

Spectrum

EF

Valence Band

Detector hf ψ

θ

e– Ekin(ψ,θ,ϕ,p)

hf

Specimen E (c)

Cu(110), θ = 0° hf = 21.2 eV ψ

EV EF

Specimen

Vacuum Level



N (E) Fermi Energy

e–

hf

(110)

3d Secondaries Ekin= 0 (eV) Ekin

14

12

EF Core level

4s 10

8

6

4

2

EB

EB Ekin= hf –  N (E)

Figure 3.8.1 (a) Experimental arrangement for photoelectron spectroscopy (see also Figure 2.6.2). The photon source can either be a UV discharge lamp, an X-ray tube, or a synchrotron storage ring. The photoelectrons are detected by using an electrostatic analyzer described in §2.6.1.2. (b) Relationship between the energy levels in the solid and the distribution of energies observed in photoemission when probed with photons of energy, hf. The kinetic energy of the photoelectrons is measured with respect to the vacuum level. Experimental measurements are often displayed in terms of the binding energy of the electrons in the solid with the Fermi level set to zero. (c) UPS spectrum (hf = 21.2 eV) from the surface of Cu(110), with normal emission. The work function ( = 4.5 eV) can also be obtained from such measurements. Adapted from [18].

Figure 3.8.1b shows the relationship for a metal between the kinetic energy distribution of the photoelectrons and their binding energy in the solid, when initially occupying the core levels or the valence band. The Fermi level, EF , is at the top of the valence band, and is separated from the vacuum level, EV , by the work function, , as shown. The binding energy, EB , is defined as EB = 0, at E = EF in solids, but in isolated free atoms or molecules, EB = 0, at E = EV ; in both cases EB < 0 (Fig. 3.7.1). However, in the photoemission literature, it is common to assign positive values for EB (which creates some difficulties for inverse photoemission, but we discuss this later), and we follow this convention

OUP CORRECTED – FINAL

190

Bonding and Spectra of Molecules and Solids for this section. Further, we can see the simple relationship between the energy levels in the solid and the kinetic energy of the photoelectrons emitted using an incident photon of energy, hf (Fig. 3.8.1b). Moreover, in practice, the variation of intensity is plotted as a function of EB (Fig. 3.8.1c). Here, to investigate the valence band of a Cu(110) surface, lower-energy (hf = 21.2 eV) UV photons are used and UPS data is acquired normal (θ = 0◦ ) to the specimen. The onset of the photoemission signal, with Ekin = hf − , is at EF , followed in terms of increasing binding energy, EB , by the shallow 4s valence band (∼2 eV wide), and then between 2–6 eV, the structured 3d band. Another important feature is the large background of secondaries, arising from those electrons that have lost some energy due to inelastic scattering processes. It is important to re-emphasize that the features seen in a photoemission spectrum are truly interpretable in terms of the electronic structure of the solid only if they emerge from the shallow escape depth (Fig. 1.3.6), of the order of ∼10 nm of the specimen where the photoemission originates without loss of energy. In practice, the surface sensitivity of XPS can be further enhanced by changing the angle of incidence of the photons (Fig. 3.8.2), which clearly shows enhanced sensitivity of the Al 2s signal to surface oxidation at shallow angles of detection (θ = 82.5◦ from the surface normal). Photoelectron spectroscopy experiments can be conveniently interpreted using a three-step model (Fig. 3.8.3). Even though photoemission takes place as a

Al 2s

Al 2s

Oxide

3

Metal

51.5°

Metal

7.5° Oxide

Intensity

5

1 Figure 3.8.2 Experiment demonstrating the surface sensitivity of XPS by changing the angle of detection relative to the surface of Al that is slightly oxidized. At 7.5◦ (θ = 82.5◦ ), the Al 2s signal is of the same magnitude for both metal and oxide, but at 51.5◦ (θ = 38.5◦ ) only the metal signal is visible. Adapted from Hüfner (2003).

120

115

120

115

Binding energy (eV) Detector Detector 7.5°

Sampling Depth e. s. d. metal

Sampling Depth 51.5° Effective Sampling Depth (e.s.d)

OUP CORRECTED – FINAL

Photoemission and Inverse Photoemission from Solids 191

(1) Photoexcitation of the electron EVB

hf

(3) Penetration through the surface

(2) Transport to the surface

EV

secondaries



EF +

EVB Valence Band

hf

crystal

vacuum

surface (a)

E Ei

E

EV

hff1

(= 0)

EF

Adapted from Hüfner (2003).

(b)

Ei

hf0



EV EF

E

Ef

E

Figure 3.8.3 Photoemission as a three-step process: (1) photoexcitation, (2) transport to the surface of the solid, and (3) penetration and escape to the vacuum.

hff0

Figure 3.8.4 Inverse photoemission (IPES) experiments can be performed in two different ways; either (a) with fixed electron energy, Ei , and tunable photon detector, or (b) with a photon detector of fixed energy, hf 0 , and variable incident electron energy, Ei . The latter (b) is also known as bremsstrahlung isochrome spectroscopy. Adapted from Hüfner (2003).

single process, for ease of understanding it can be divided into three steps: (1) photoexcitation of electrons, (2) transport of the electrons to the surface with simultaneous production of secondaries, and (3) penetration through the surface layer and escape into the vacuum, where it is collected by the spectrometer. Inverse photoemission spectroscopy (IPES) measurements (Fig. 3.7.1c) can be carried out in two different ways (Fig. 3.8.4a,b). In the first method (a),

OUP CORRECTED – FINAL

192

Bonding and Spectra of Molecules and Solids the specimen is probed with electrons of a fixed energy, Ei , and the range of emitted photons, produced by the bremsstrahlung process, as the electron is rapidly brought to a stop inside the material, is measured using a tunable X-ray detector. For example, photons measured with energy, hf1 , as shown, correspond to an unoccupied level, E1u =  − (hf1 − Ei ), above E F . The alternative method of bremsstrahlung isochrome spectroscopy (BIS) (Fig. 3.8.4b) probes the surface with electrons of variable energy, Ei + Ei , and the photons of a specific energy, hf 0 , are detected. The corresponding energy of the unoccupied level would be E1u = Ei (also, above E F ). Note that in PES, the energy range between EF and EV is not accessible, but in IPES/BIS the inaccessible range is for energies less than EF , and it is the unoccupied states, EF < E < EV , that are truly resolved. In other words, the two methods are complementary (Fig. 3.8.5). We can compare the photoemission spectra, using UV light, of the same hydrocarbon material in the gas and solid phases (Fig. 3.8.6) for benzene (C6 H6 ). In the gas phase (a), the photoemission describes the occupied molecular orbitals (MO), starting with the top of the filled π MO as a band around EB ∼ 9 eV. The higher energy bands reflect further changes in molecular geometries as the electrons are removed from different MOs. Overall, the energy levels in the molecular solid, (b), closely reflect the energy levels of the MOs in the individual molecules of C6 H6 . However, the bands appear broader in the solid compared to the gas; this is attributed to the crystal vibration modes that are likely excited in the photoionization process. Moreover, the energy bands are shifted to lower energies in the solid by ∼1.15 eV. This shift, as discussed earlier (§2.6.3), is the

Fe

Intensity

Co

Figure 3.8.5 X-ray photoelectron spectra [8], continuous line, and inverse photoemission [9], dashed line, obtained at a photon energy of 530 eV, for the 3d band in the transition elements, Fe, Co, Ni, and Cu. Notice the complementarity of the two techniques.

Ni

Cu

Binding energy (eV) (PES)

10

0 EF

10

Energy above Fermi level (eV) (Inverse PES)

OUP CORRECTED – FINAL

Photoemission and Inverse Photoemission from Solids 193 (a) C2H6 (Gas)

8 (b)

8

12

16 20 C2H6 (Solid)

12 16 Electron binding energy (eV)

20

result of the response of the other electrons in the solid to the ionization process. The positive hole left behind at the site of ionization will instantaneously polarize the electrons in the surrounding molecules, which will affect the energy required to produce the hole! Such polarization energy has been calculated (see Cox, 1987) to be of the order of 1 eV, in agreement with the experiment. The effect of ligands in photoemission studies can be understood further by considering the XPS spectra in three different copper dihalides (Fig. 3.8.7). In the ground state, these three dihalides have Cu in the 3d 9 configuration and a filled ligand shell, L, which is 2p for F – , 3p for Cl– , and 4p for Br – . Thus, the electronic configuration for dihalide can be written as 3d 9 L. The photoexcitation of the Cu 2p levels can lead to two possible final states: either the 3d 9 L configuration remains unchanged, or after the core hole is created, one electron is transferred from the ligand shell to the Cu d-shell, resulting in a final 3d 10 L –1 configuration. If we now recall the dipole selection rules (§2.3.1) and the spin-orbit coupling (§2.2.2), we can assign the main peaks observed in the XPS spectrum (Fig. 10 −1 and 2p−1 3d 10 L −1 . The two satellites are 2p−1 3d 9 L and 3.8.7) as 2p−1 3/2 3d L 1/2 3/2 9 2p−1 1/2 3d L, respectively. In this first approximation, the main line energy shows a strong correlation with the ligand as the photoionization energy includes the energy required to produce a hole in the ligand valence orbital, which in turn depends on the nature of the ligand, and explains the observed shift in the position of the main lines. Figure 3.8.8a shows the UPS measurement for polycrystalline Ag at room temperature, using hf = 21.2 eV incidence, for the energy region around EF (compare with Figure 3.2.3b). The instrument used had an energy resolution of 25 meV and the fit of the data to the Fermi–Dirac distribution (3.2.2) is in good agreement. The width of the Fermi–Dirac distribution is ∼100 meV or 4kB T, which is larger than the resolution, and determines the shape of the curve. Measurements around the Fermi “edge” at lower temperatures are a good way to determine the energy

Figure 3.8.6 UPS spectra of C6 H6 (benzene) in (a) the gas phase, and (b) as a molecular solid. Other than the broadening and a relaxation shift of ∼1.15 eV, in the solid the two spectra are not very different. Adapted from [10].

OUP CORRECTED – FINAL

194

Bonding and Spectra of Molecules and Solids

CuF2

main line

main line satellite

satellite

CuCl2

Spin orbit splitting Spin orbit splitting

Figure 3.8.7 XPS spectra of copper dihalides, CuF2 , CuCl2 , and CuBr2 . The main lines are assigned 10 −1 and 2p−1 3d 10 L −1 , as 2p−1 3/2 3d L 1/2 respectively, and their energies vary because of the difference in the binding energies of the ligands. Adapted from [11].

CuBr2

–1 3d10L–1 2p3/2

930

–1 3d 9L 2p–1 3d10L–1 2p–1 3d 9L 2p3/2 1/2 1/2

950 Binding energy (eV)

970

resolution of an UPS instrument. Figure 3.8.8b illustrates a similar measurement using a very high-resolution spectrometer for polycrystalline Ag at T = 8 K. As we can see, using a 3 meV Gaussian resolution function to convolute the Fermi–Dirac distribution, gives a very good fit to the experimental data. Figure 3.8.9 compares XPS and UPS measurements of a complex oxide, Na0.7 WO3 . In XPS, the X-ray photons used have sufficient energy to ionize the core levels, including the Na1s, O1s, W4d, and W4f levels. Additional Auger transitions can also be observed. The small shift in binding energies, from that of the pure elements, is consistent with the oxidation states. The valence energy levels are better probed by UPS, because the inherently smaller energy line widths, lead to better energy resolution. The valence spectra show two bands, assigned to the W5d and O2p levels, and the relative intensities of the two bands is consistent

OUP CORRECTED – FINAL

Photoemission and Inverse Photoemission from Solids 195 (b) 3000

(a) Ag

1

=

E – EF

e

kBT

hf = 21.2 eV T = 8K Counts/channel

Counts/channel

1200

800

Ag

hf = 21.2 eV T = 300K

+1

400

2000

1000

Experimental data Gaussian function (FWHM = 3 meV)

FDF convoluted with Gaussian

Fermi-Dirac function

100 meV 0 0.4

0.2

0 –0.2 Binding energy, EB (eV)

0 10

–0.4

5

0

–5

–10

Binding energy, EB (meV)

Figure 3.8.8 UPS spectrum (hf = 21.2 eV) of polycrystalline Ag at (a) T = 300 K and FWHM 25 meV, and (b) T = 8 K and FWHM ∼3 meV, showing the energy distribution around the Fermi level. As expected, the data fits very well to the Fermi–Dirac distribution. Adapted from Hüfner (2003).

O 1s

(a) W 4f

W 4d

0

250

Na 1s

(b) W 5d

750 500 Binding energy (eV)

1000

1250

W O 2p 5d × 10

O 2p

0 5 10 Binding energy (eV)

(c)

0 5 10 Binding energy (eV)

Figure 3.8.9 PES spectra of Na0.7 WO3 . (a) A wide scan showing the core levels. (b) XPS of the valence band and (c) a higher resolution UPS of the valence band. Adapted from Cox (1987).

with the difference in their ionization cross-sections, as a function of the incident photon energy. This section ends with a discussion of a practical application of photoemission spectroscopy in materials research [12]. CaF2 has a crystal lattice structure that matches very well to Si(111), and together they are a model system to study insulator–semiconductor epitaxial6 behavior by photoemission spectroscopy. Moreover, the PES can be acquired in situ, if the spectrometer is combined with a UHV growth chamber, to monitor the evolution of the electronic structure as a function of epitaxial growth. Figure 3.8.10b shows three such PES spectra corresponding to different stages of the growth of CaF2 ; starting with a

6 Artificial growth of crystals on a crystalline substrate with a well-defined crystallographic orientation relationship between them.

OUP CORRECTED – FINAL

196

Bonding and Spectra of Molecules and Solids Fluorine atom hf (135 eV)

CaF2 on Si(111)

e–(124 eV)

F 2s

Bulk Ca 3p

F 2p

Thick film

Figure 3.8.10 Practical application of PES in materials research. Here it is used to study the evolution of the bonding of CaF2 (insulator) to a Si(111) surface. The energy levels (left) and three PES spectra for a monolayer (bottom), a thin 1.1 nm layer (middle), and a thick 5 nm film (top), all of CaF2 , are shown. Further details are discussed in the text. Adapted from [12].

F 2p F 2s

Vacuum level CaF2 valence band Core levels

Intensity

e–(103 eV) 0

2.3 eV

F 2p

Thin film

Interface Ca 3p 0 Ca 3p

F 1s 0

F 2p Monolayer

30 20 10 Energy below EVB (eV)

0

single monolayer or a molecule thick film of CaF2 (bottom), to a relatively thick (∼5 nm) film (top), and a thin (∼1.1 nm) film in between. Incident photon energy was 135 eV, and the experiments were performed using a synchrotron source. The plots show all energies in eV below the top of the valence band. In the monolayer spectrum, one observes narrow atom-like sharp features at 8.3 eV (F 2p) and 25.6 eV (Ca 3p). In the case of Ca, this feature is associated with bonding to the surface. The role of fluorine is minimal, as the Ca–Si bonding is preferred, and Ca retains a larger fraction of the shared electron, as reflected in the lower binding energy. In the thin film spectrum, the F 2p peak broadens and has characteristics of the bulk valence band; in addition, the Ca 3p now develops a split peak, at 25.6 eV and 27.9 eV, with features of the bulk and surface valence bands. One can assign the 2.3 eV difference to the binding energy of Ca in the bulk and surface environments. In addition, a core F 2s feature is also seen. Finally, in thick films, both F 2p and Ca 3p, show fully developed bulk features; now, the Ca outer 4s2 electrons are strongly attracted by the surrounding eight highly electronegative fluorine atoms, and thus increasing the binding energy of the remaining 3p electrons. It is clear from this discussion that PES has great utility in chemical identification and analysis, as discussed in §2, and also in elucidating bonding in different chemical environments. The only requirement is a specimen with a clean surface that is representative of the bulk structure. The examples of PES in this section are all angle-integrated spectra; by this we mean that electrons emitted over a wide solid angle are collected and displayed. In practice, the energy distribution of electrons photoemitted from a solid is dependent on the direction of both the incoming photon and the outgoing photoelectron. If data were to be acquired

OUP CORRECTED – FINAL

Absorption Spectroscopies–Probing Unoccupied States 197 in an angle-resolved manner, detailed information on the 3D band structure (Fig. 3.3.3) of the solid or the 2D band structure of a surface adsorbate layer, including the orientation of such adsorbed molecules, can be obtained. In practice, to get a full appreciation of the full range of capabilities of this technique, it requires variation of the photon energy (using synchrotron radiation) and the polarization of the incident radiation. Further details of PES can be found in many texts and monographs, including Hüfner (2003), emphasizing fundamentals, and Watts (1990) for a good introduction from a materials applications point of view. Example 3.8.1: For Al Kα radiation (1,486.7 eV) incident on a Cu film, determine the ratio of 2s to 3s photoelectron yields based only on crosssections (Fig. 2.6.17) and escape depths (Fig. 1.3.6). Solution: The probability, Pie , per incident photon to create a photoelectron, simplified (2.6.4), is given by Pie = σ k Nt, where Nt is the number of atoms/cm2 in a layer of thickness t, and σ k is the cross-section (Fig. 2.6.17) for ejecting a photoelectron from the k-orbital. However, the number of electrons that can escape from the solid and be detected decreases with depth as exp(–x/λ), where λ is the mean free path length (Fig. 1.3.6). Thus, the probability, Pd , of an incident photon producing a detectable photoelectron is then Pd = σ k Nλ. We ignore instrument efficiencies of detection. From Table 2.2.2, the binding energies for Cu, Eb2s = 1096.7 eV , Eb3s = 122.5 eV From Figure 2.6.17, the corresponding cross-sections for Al Kα radiation are σ 2s = 8 × 104 , σ 3s = 1.5 × 104 . 2s = For incident Al Kα radiation, the photoelectrons have energies EPE 3s 1486.7 − 1096.7 = 390 eV and EPE = 1486.7 − 122.5 = 1364.2 eV . The corresponding mean free path lengths (Fig. 1.3.6) are λ2s = 10 and λ3s = 20 Å. 2s 2s P 2s 8×104 N10 Thus d3s = σσ 3s Nλ = 1.5×10 4 N20 ≈ 2.67 Nλ3s Pd

3.9 Absorption Spectroscopies–Probing Unoccupied States 3.9.1

X-Ray Absorption Spectroscopy (XAS)

Figure 3.7.1 summarizes electronic transitions that form the basis of experimental studies of the electronic structure of both occupied and unoccupied states in materials. Section 3.8 described photoemission spectroscopy (PES) techniques and discussed how XAS complements PES methods (Fig. 3.7.2). In XAS, the lowest

OUP CORRECTED – FINAL

Bonding and Spectra of Molecules and Solids

Absorption

198

Figure 3.9.1 X-ray absorption L-edge spectra of tetrahedrally coordinated silicon in (a) gas phase SiF4 , and (b) solid SiO2 . Adapted from [13].

7 An exciton is an electrically neutral quasiparticle arising from the bound state of an electron and a hole, attracted to each other by the electrostatic Coulomb force. It typically exists in insulators, semiconductors, or in some liquids.

SiF4

(a) SiO2 100

120 140 Photon energy (eV)

(b) 160

energy excitations are from occupied to unoccupied states in the valence band; here the shape of the spectrum is a convolution of the single particle densities of states of the initial and final levels, and since its interpretation is complicated, it is not popularly used. However, higher-energy excitations correspond to local transitions from a core level to an unoccupied state in the valence band, and subject to dipole selection rules, (2.3.2), as discussed in §2.3.1. The interpretation of the spectra, in most cases, is simple as it relates directly to the final unoccupied density of states. Such a single-electron transfer picture of XAS gives data that is related to the specific site and symmetry-selected density of states. To illustrate this point, consider the Si-L edge absorption spectra (Fig. 3.9.1) for a gas (SiF4 ) and a solid (SiO2 , quartz); in each case, four highly electronegative atoms tetrahedrally coordinate the silicon. The two spectra are very similar as they reflect transitions of electrons from the Si 2p levels to unoccupied states, confirming the importance of local symmetry in the XAS spectra of both gas and solid phases. Note that, at this introductory level, we ignore the interaction between the excited electron and the core hole, the possible formation of excitons,7 and the broadening of the XAS spectra because of the short lifetime (Fig. 2.3.1) of the core hole due to their decay either by fluorescence or Auger electron emission. Furthermore, XAS is similar to inverse photoemission, introduced in the last section, but with one difference. Unlike XAS, inverse photoemission samples all the unoccupied states and is not limited by the dipole selection rules. In practice, XAS studies are best carried out using synchrotron radiation, where monochromatic X-rays suitable for XAS, with energies >800 eV, are generated by Bragg diffraction using a double-crystal monochromator (§7.9.5.2). However, for lower-energy (50–800 eV) X-rays, diffraction gratings (Fig. 3.5.1) are preferred. The absorption of monochromatic X-rays can be measured in two different ways (Fig. 3.9.2). In the transmission mode (a), the spectrum is the ratio of the beam intensity before and after it passes through the specimen. In general,

OUP CORRECTED – FINAL

Absorption Spectroscopies–Probing Unoccupied States 199 (a) Transmission Detector

hf

(c)

100

0 (b) Photoyield

EXAFS

Absorption

Transmission %

Specimen

EV EF

UDOS

Specimen

DOS

hf NEXAFS (XANES)

Photoyield %

e–e– Detector 100

hf EB

12.9

13.1

13.3 13.5 Photon energy (keV)

13.7

13.9

0

Figure 3.9.2 Schematic diagram of XAS measurements in (a) transmission mode through a thin film specimen, and in (b) the photo-yield mode. (c) A typical absorption L 3 edge of Pb in an oxide specimen, showing energy ranges of interest for NEXAFS and EXAFS. The inset shows an idealized energy-level diagram illustrating how the transitions to levels in the unoccupied density of states, subject to dipole selection rules, contribute to the near-edge structure. this method is used to study deeply bound core levels, but for energies 10 nm) sensitivity than PES (0.5–2 nm), because all electrons, including those that have suffered inelastic scattering events before escaping the specimen are collected and included in the

OUP CORRECTED – FINAL

200

Bonding and Spectra of Molecules and Solids XAS signal. Recent developments in transmission electron microscopy (TEM) hardware, including field emission sources and monochromators, make it possible to obtain measurements equivalent to XAS with comparable energy resolution using electron energy-loss spectroscopy (EELS), but with much better spatial resolution (∼1 nm). The underlying physical principles of XAS and transmission EELS in the forward scattering direction are the same, and we discuss this further in §9.4.

3.9.2 Near-Edge and Extended X-Ray Absorption Fine Structure (NEXAFS and EXAFS) Figure 3.9.2c shows a typical X-ray absorption spectrum of the L 3 edge of Pb from a solid specimen of BaPb1-x Bix O3 , illustrating two types of fine structure superimposed on the generic X-ray absorption edge profile (§2.4.2). The first one, typically extending ∼50 eV above the edge, referred to either as near edge X-ray absorption fine structure (NEXAFS), or as XANES, is determined by the final density of states, the transition probability, and many body effects. The analysis of NEXAFS is quite complex, often requiring substantial theoretical effort (see Fuggle and Inglesfield, 1992). Therefore, rather than discuss this complicated topic, we restrict our discussion to the example of the “white lines” observed in 3d transition metals (Fig. 3.9.3) where the interpretation of the fine structure can be based on a simple one-electron picture. In addition to the near-edge structure, a second fine structure extending from ∼50 eV above the absorption edge for several hundreds of eV, due to the interference effects of the wave function of the excited electrons propagating in the solid, is also observed. This is called the extended X-ray absorption fine structure (EXAFS). After a brief introduction to the white line transitions observed in NEXAFS, we conclude this chapter with a discussion of EXAFS.

Figure 3.9.3 X-ray absorption spectra, recorded by total electron yield detection, of the L 3 and L 2 edges of transition metals (Fe, Co, Ni, and Cu) with the gradual filling of their d-orbitals. Note that the white lines are absent in Cu because of its filled d shell. Adapted from Schlachter and Wuilleumier (1994).

Total electron yield

10

Fe

Co

Ni

Cu

5

0 700

750

800 850 900 Photon energy (eV)

950

OUP CORRECTED – FINAL

Absorption Spectroscopies–Probing Unoccupied States 201 Figure 3.9.3 shows L-edge ( p → d) XAS spectra, satisfying dipole selection rules and obtained by total electron yield, for the ferromagnetic 3d transition metals (Fe, Co, Ni, and Cu). We observe separate L 3 (2p3/2 → 3d unoccupied ) and L 2 (2p1/2 → 3d unoccupied ) transition features, because of the strong spin-orbit splitting of the 2p levels (§2.2) in the edges for all elements except Cu (absent, because of the 3d 10 configuration). The p → d transition, which is divided into three different cases depending on how the d valence states are treated, can be understood in the one-electron picture by referring to Figure 3.9.4. In the first case, all 10 d states are assumed to be degenerate, with one of them being empty, and an electron is excited from either the 2p3/2 or 2p1/2 spin-orbit split core states into the d-hole, giving rise to the sharp L 3 and L 2 features in the XAS spectra. In the second case, we have assumed that the d states are also split into spin-up and spin-down states as a result of exchange interactions in the ferromagnetic elements. If an external magnetic field, H, is applied such that the spin-up states are lower in energy, but with a hole in the spin-down state that can accommodate a photoelectron, then the transition is allowed, as shown in (b). The third case (c) corresponds to the spin-orbit splitting of both the final 3d and initial 2p states. Now, the selection rule that applies is j = 0, ±1, and the 1/2 → 5/2 transition is not allowed. As a result of these various factors, in general, the ratio of the L 3 /L 2 white line intensity ratio can change dramatically along the 3d transition metal series (Fig. 3.9.5). Further, the One-electron picture of p→d transition (a)

(b)

(c) d5/2

d

d3/2 L3

L2 forbidden

p3/2 p1/2 Exchange

Spin-orbit

Figure 3.9.4 One-electron picture of white line transitions, 2p → 3d, in transition metals for three different consideration. (a) Spin-orbit interaction only in the 2p core level. (b) In addition to (a), the exchange interactions in the d shell, causing a splitting of the spin-up and spin-down states for the ferromagnetic elements is included. (c) The exchange interaction in the d shell is replaced by the spin-orbit interaction with splitting into d 5/2 and d 3/2 levels. Adapted from Schlachter and Wuilleumier (1994).

OUP CORRECTED – FINAL

Bonding and Spectra of Molecules and Solids

L3

L3 L3

Figure 3.9.5 The ratio of the white line intensity, L 3 /L 2 , varies linearly with atomic number along the 3d transition metal series. Adapted from electron energy-loss measurements [14]; note that the photon energy is not to scale. Inset shows that the total white line intensity is linear with the number of d-holes. Adapted from [15].

Intensity

L3

L2

L2

L3

Ni

L2

L3 L2

Cu

L2

Co Fe

L2

Cr V Ti

White line intensity (Mb2eV)

202

100 80 Fe Co 60 40 Ni 20 Cu 0 0 2 1 3 4 Number of d-holes

Photon energy

sum of the intensities of L 3 and L 2 white lines is proportional to the number of d-holes (Fig. 3.9.5, inset) and the white line ratio is also sensitive to the oxidation state of the transition metal (see §9.4). The second, long-range oscillations (EXAFS) (Fig. 3.9.2c) have been developed into a useful, non-destructive, element-specific technique to directly probe the atomic environment of a particular X-ray absorber atom. It provides chemical bonding information (nearest neighbor distances and coordination number) for both crystalline and amorphous materials, and also for nonsolids such as molecules, surface adsorption species, etc. In principle, it complements diffraction (to be introduced in Chapter 4), as a tool for structural analysis. The basic principles of EXAFS, using either inner (K, L) or outer (M, N) shell ionization, are quite easy to understand, even though its practical implementation requires sophisticated data analysis. As the X-ray photon energy approaches the binding energy characteristic of a core level of an element, absorption occurs and a sharp onset of the absorption edge is observed. As the photon energy is increased above the absorption edge, the core electron has sufficient energy to be excited to the continuum. The ionized photoelectron propagates as a wave (recall the waveparticle duality discussed in §1.3.3) and is scattered by the neighboring atoms (Fig. 3.9.6a). The outgoing wave from the absorber atom and the back-scattered wave from the neighboring atoms interfere, both constructively and destructively, to give the modulated EXAFS spectrum. The EXAFS spectrum depends on the local atomic structure distribution in the immediate environment of the absorber atom (Fig. 3.9.6b); in particular, the frequency of EXAFS oscillations depends on the distances between the absorber and backscattering atoms, and its amplitude depends on the number, type, and

OUP CORRECTED – FINAL

Absorption Spectroscopies–Probing Unoccupied States 203 Neighboring atom

(a)

(b)

R X-ray absorbing atom

R1

Outgoing photoelectron wave (i)

R2 Backscattered photoelectron waves

(ii) R R

(iii) Backscattered photoelectron waves

Photon energy

Figure 3.9.6 (a) A schematic illustration of the phenomenon of EXAFS. Top: The outgoing photoelectron wave from the absorber atom encounters a nearest neighbor atom, at a distance, R, in the crystal. Middle: Destructive interference of the backscattered and outgoing photoelectron waves. Bottom: Constructive interference at the site of the absorber atom. (b) The effect of nearest neighbor distance, (i) and (ii), and coordination number, (iii), on EXAFS. The frequency of the oscillations is related to the distance, R; when R2 < R1 , we observe a longer EXAFS period. The amplitude of EXAFS is related to the coordination number; when the number of backscattering atoms increases, the larger is the EXAFS amplitude (here, 6 and 2 nearest neighbors are compared). Adapted from Stohr (1981).

arrangement of back-scattering atoms. It is also possible to identify the backscattering atoms chemically, as they have a unique way of altering the phase of the back-scattered wave. The analysis of EXAFS data can be understood following Figure 3.9.7, which shows the absorption spectrum for the Mo-K edge. First the shape of the edge is fitted to a polynomial function (ln I 0 /I t versus E, in Figure 3.9.7a) to obtain the modulatory function, μ(E). The smooth absorption background function,

OUP CORRECTED – FINAL

Bonding and Spectra of Molecules and Solids (a)

Constructive interference

Absorp

Destructive

tion sp

ectrum

Polyn Edge

ln (I0/It)

omial

Backg

round

fit, μ

, μ0

EXAFS

XANES

Pre-edge 19.8

20.0

20.2

20.4

20.6 20.8 Photon energy, keV

21.0

21.2

21.4

21.6

(c)

28

Fourier transform intensity

(b)

14 k3χ (k)

204

0 –14 –28 0

4

8

12 k

(Å–1)

16

20

0

1

2

3

4

5

6

7

R´ (Å)

Figure 3.9.7 (a) Mo K-edge absorption spectrum obtained by transmission through a thin metal foil at 77 K with synchrotron radiation. The fitted polynomial modulatory function, μ(E), and the smooth background function, μ0 (E), both shifted vertically for clarity are also shown. (b) Background subtracted and k3 weighted EXAFS spectrum, χ (k)k3 versus k, based on the Mo K-edge data in (a). (c) Fourier transform of (b), as a function of the effective interatomic separation, R’, observed; this does not include the phase shifts. Two large peaks, observed at 2.38 Å and 2.78 Å, and two small peaks observed at 4.04 Å and 4.77 Å, are consistent with interatomic spacing in Mo, provided a phase shift of 0.37 ± 0.2 Å is introduced for all backscattering events.

OUP CORRECTED – FINAL

Absorption Spectroscopies–Probing Unoccupied States 205 μ0 (E), is then subtracted from μ(E) to give the EXAFS oscillations as a function of energy χ (E) = (μ(E) − μ0 (E)) /μ0 (E)

(3.9.1)

The energy (eV) is now converted to a wave-vector (k) scale, where √ k = 2π

2m (E − E0 ) = 0.263(E − E0 )1/2 h

(3.9.2)

to give χ(k), the oscillating portion of the absorption coefficient as a function of the wave-vector. Note that E 0 is the energy threshold at which the onset of the K-edge is observed (for Mo-K it is 20,002 eV), chosen to define the origin of the EXAFS spectrum. In other words, k = 0, when the incident X-ray photons have energy E = E0 , and the photoelectrons have no kinetic energy. Figure 3.9.7b shows EXAFS data, χ (k), which are then normalized by multiplying by kn to compensate for amplitude attenuation as function of k (the value of n = 1, 2, or 3, and is determined empirically, based on best fit to the data), and χ (k) k3 , for the Mo-K edge. Finally, the data is Fourier transformed to provide a pseudo-radial distribution function, R  , of the atoms around the absorber atom (Fig. 3.9.7c). In general, when compared to the true distances in the solid, the peaks observed in the Fourier transform are shifted by a small distance. This is because of the element specific phase shift introduced in the backscattering process. For the case of Mo, the observed distances, R  , given by the peaks in Figure 3.9.7c, are at 2.38 Å, 2.78 Å, 4.04 Å, and 4.71 Å. The actual Mo–Mo distances, R, in the metal are 2.73 Å, 3.15 Å, 4.45 Å, and 5.22 Å. Applying a uniformly small phase shift of 0.37 ± 0.2 Å gives a good fit between experiment and all true nearest neighbor distances. An important example of the application of XAS is in understanding the photosynthesis reaction [16]. It helps establish the relationship between structure and function of metalloproteins, such as Mn, Ca, and possibly Cl, which are the metal sites in the oxygen-evolving complex (OEC) involved in photosynthesis. In particular, the role of the Mn4 Ca cluster in photosynthesis (PS II) is of special interest. Figure 3.9.8 shows the Mn K-edge from spinach PS II, highlighting the EXAFS region of the spectrum. Following the analysis described in Figure 3.9.7, the first observed oscillatory EXAFS contribution, k3 χ (k), is shown in Figure 3.9.8a (inset), which is then Fourier transformed to obtain the apparent distances (Fig. 3.9.8b) between the atoms of interest. The EXAFS is interpreted in real space as shells at 1.8 Å (Peak I) attributed to O or N atoms, and a second shell at 2.7 Å (Peak II) attributed to Mn–Mn interactions. Further, an additional shell for Mn and Ca is also observed at ∼3.35 Å (Peak III). By tracking the EXAFS changes [16] as a function of the photosynthesis reaction, such measurements showed that the Mn4 Ca complex (OEC) undergoes structural changes that are triggered by the changes in oxidation state and protonation or deprotonation events.

OUP CORRECTED – FINAL

206

Bonding and Spectra of Molecules and Solids EXAFS

(a)

(b)

I

II

O Mn

FT Amplitude

I/I0

χ(k)k

S1 1.8 Å (I)

3

O Mn

Mn

4 6 8 10 Photoelectron Wavevector, k(1/Å) 6600

6700 6800 Energy (eV)

6900

7000

0

2

Mn 3.3 Å (III) O

III

6500

Mn O 2.7 Å (II)

4

Ca 3.4 Å (III)

6

8

Apparent distance (Å)

Figure 3.9.8 (a) Mn K-edge EXAFS spectrum from a spinach specimen used for studying the role of metal clusters in photosynthesis. The inset shows the spectrum in k space. (b) The Fourier transform of the k space EXAFS spectrum, gives the inter atomic distance of the other elements (Ca and O) with respect to Mn. Adapted from [16].

Further details of the use of XAS in understanding photosynthesis can be found in [16]. It is clear from this example that EXAFS also provides useful information to biochemists and structural biologists. In particular, it provides element-specific structural information (distance between absorber and back-scatter atoms up to a distance of ∼5 Å, with a precision of 0.02 Å) without interference from a periodic arrangement of atoms, that is independent of the state of the sample (the sample can be a powder, a solution, or specifically a frozen solution for biological materials), and at intermediate stages of a chemical process. In materials science, EXAFS is a versatile method that complements X-ray diffraction with unique advantages in studying in situ processes on materials surfaces, such as catalysis and electrochemical reactions. Further details can be found in specialized texts and monographs; for example, see Teo and Joy (1981).

3.10 Select Applications 3.10.1

Structure of Proteins Resolved by FTIR

We can generalize the result of Example 3.4.1 and state that in a protein if any specific atom is replaced by its isotope, the vibrational frequency of that specific bond will change proportional to 1/(μ)1/2 , where μ is the reduced mass of the two

OUP CORRECTED – FINAL

Select Applications 207 atoms involved in the bond. Thus, by substituting 1 H in proteins with deuterium, the structure of proteins can be resolved [19]. Further, in these experiments [19] the specific carbonyl group involved was substituted with 13 C and 18 O, such that the specific 13 C=18 O bond will give a specific signature (wavenumber) different from other C=O bonds. The specific question addressed was the hydrogen bond between different amide groups in the protein (Fig. 3.10.1a–d). On deuterating the protein, the IR peak shifted to lower wavenumbers as the effective mass increased. By comparing with density functional theory calculations, it was found that the shift in the IR absorption peak to lower wavenumbers (Fig. 3.10.1e) for the 13 C=18 O bond was caused by the deuteration of positions (B) shown in red, but not in positions (A) shown in blue. In other words, noncovalent bonds, e.g. the protons in positions A, do not affect the C=O bonding, but the covalent bond (position B) influences the vibration frequency. Furthermore, they also analyzed the in-plane bending of the N–H bond at position B, under the influence of an external electric field; the interested reader is referred to [19] and the related supplementary information.

(a)

(b)

(e) O

O CH3

C N H

CH3

N H

13C

CH3 CH3

(c)

N D

13C

CH3

(d)

O

O CH3

C N D

CH3

CH3

N H

CH3

10 cm–1

0.8 0.6 0.4 0.2 0 1610 1600 1590 1580 1570 Wavenumber / cm–1

18O

18O

CH3

C N H

1.0

CH3

18O

18O

CH3

C N D

Absorbance (a.u.)

CH3

13C

CH3 CH3

N D

13C

CH3

Figure 3.10.1 (a–d) Possible arrangement for the deuteration of the protein. Each system has two amide carbonyl groups, with the isotopically labeled one, participating in the hydrogen bond, shown in green. Further the hydrogen-bonding hydrogen (Position B) is shown in blue, and the amide hydrogen (Position A) is shown in red. (e) shows the change in IR absorption between (a) and (b). Adapted from [19].

OUP CORRECTED – FINAL

208 Bonding and Spectra of Molecules and Solids

3.10.2 Analysis of Catalytic Particles by XAS and XPS The structure, chemistry, and oxidation states of catalytic particles are key to their functionality. To purify automobile exhaust, Pt-based three-way catalyst (TWC) nanoparticles are used. However, above 800◦ C, these Pt nanoparticles sinter, decreasing their surface area available for reactions in oxidative environments. XAS was used to study mechanisms to inhibit such sintering [20]. Pt/cereia-based catalysts (referred to as CZY) did not sinter at 800◦ C, but Pt/Al2 O3 did. The Pt L3 edge was studied, and both near-edge structure (XANES) and extended fine structure were analyzed. The white lines observed at the onset of the Pt L3 edge (Fig. 3.10.2a) reflects the vacancy of the 5d orbitals of Pt, and its intensity can be correlated with the Pt oxidation state. Thus, we can conclude that Pt/Al2 O3 , with white line features similar to Pt foil, is in the Pt0 (metallic) state after aging. On the other hand, Pt/CZY, shows white line features similar to PtO2 , thus indicating that Pt2+ and Pt4+ species were present in these aged samples. If the white line intensity is linear with oxidation state, the aged Pt/CZY catalysts can be expected to have an average oxidation state of 3.53. Figure 3.10.2b shows EXAFS data, after Fourier transformation, for the same set of samples. The peak at 2.76 Å for the Pt foil is assigned to the Pt–Pt bond, and the peaks at 2.04 Å and 3.10 Å for the PtO2 powder are assigned to the Pt–O and Pt–O–Pt bonds, respectively. Again, the Pt/Al2 O3 data (FT spectrum) agrees with that of the Pt foil. On the other hand, the aged Pt/CZY specimen shows data different from both Pt foil and PtO2 powder. The peak at 2.02 Å is close to the Pt–O bond length in PtO2 powder, but the peak at 3.01 could only be fit to Pt–O–Ce bond after considerable simulation [20], giving an indication of how/why sintering is inhibited in Pt/CZY catalysts. In other metal catalysts it is important to maintain the metal oxidation state, even upon exposure to air, as the catalyst facilitates redox reactions. XPS was used to analyze the stability of Co(Ni)MoS catalysts before and after exposure to air for three minutes (Fig. 3.10.2c,d). It is clear from the figure(s) that the relative amount of CoMoS decreased, while the oxide phases, Co9 S8 , and Co(II)-oxide increased. Further details can be found in the original manuscript [21].

Summary When atoms form molecules and solids, their electronic structure are significantly altered, particularly the energy levels of those outer electrons involved in the bonding. In addition, the vibrational and rotational degrees of freedom in molecules modulate their electronic structure. As a result, molecules and polymers exhibit a combination of electronic (2–10 eV, UV-Vis), vibrational (10-2 –2 eV, IR), and rotational (10–5 –10–3 eV, microwave) energy levels that can be probed by appropriate spectroscopy methods.

OUP CORRECTED – FINAL

Summary 209 (a)

(b) 80

Pt-Pt

Pt foil

4

Magnitude of FT/Å

Normalized XANES

5

Pt/Al2O3

3

Pt/CZY

2

PtO2 1

Pt foil

60

Pt/Al2O3

40 Pt-O Pt-O-Ce

Pt/CZY 20

Pt-O-Pt

PtO2 0 11.54

11.58 Photon energy /keV

0 0

11.62

1

2

3

4

5

6

R/Å (d)

(c) Phase

Relative Amount of Co in Phase

Co8S9

13.40

CoMoS

62.24

Co(II) oxide

24.36

Co2p3/2scan CoMoS

Phase

Relative Amount of Co in Phase

Co8S9

19.01

CoMoS

43.01

Co(II) oxide

37.97

Co2p3/2scan

CoMoS

Co(II) oxide Co(II) oxide

790

786 782 778 Binding energy (eV)

Co8S9

Co8S9

774

790

786

782

778

774

Binding energy (eV)

Figure 3.10.2 (a) Pt L3 XANES and (b) Fourier transformed k3 x data for Pt L3 EXAFS, for Pt supported catalysts aged at 800◦ C in air, together with standards for Pt metal and PtO2 . Co 2p3/2 XPS spectrum of (c) fresh and (d) exposed to air, catalyst of CoMoS particles. Note that the Co 2p3/2 peak was deconvoluted into component peaks, which requires some prior knowledge of the composition of the specimen. (a, b) Adapted from [20]. (c, d) Adapted from [21].

OUP CORRECTED – FINAL

210 Bonding and Spectra of Molecules and Solids Light, especially in the IR region, incident on a molecule or molecular solid can be either absorbed or scattered. The scattering can either be elastic (Rayleigh scattering) or inelastic (Raman scattering), the latter resulting in a signal that exhibits either a smaller (Stokes scattering) or larger (anti-Stokes scattering) frequency. Unlike the single photon involved in an IR absorption measurement, Raman scattering is a two-photon process. First, the probing photon is absorbed raising the molecule to a virtual excited state; this is followed by a decay and emission of a second photon of either the same (Rayleigh) or altered (Raman) frequency. The change in frequency arises primarily from the internal modes of quantized vibration and secondarily from the rotation of the molecule. Further, IR absorption and Raman scattering are complementary in nature. For a given molecule, if the derivative of its dipole moment with respect to the vibration mode is nonzero, it can show IR activity; alternatively, if the derivative of its polarizability with respect to the vibration mode is nonzero, it can show Raman activity. Signals in the IR region of the electromagnetic spectrum can be detected by dispersion using gratings or interferometry methods. The latter detects all wavelengths at the same time (parallel detection) with much higher signal collection speeds. These two spectroscopy methods find wide use in a range of technological applications involving polymers, and include the ability to monitor the curing of fiber-reinforced polymer composites increasingly used in the manufacture of fuselages of commercial airplanes. A different class of spectroscopy methods provides details of the electronic structure of a crystalline solid using either X-rays or electrons as probes. In particular, they include X-ray or UV photoemission (PES, probing occupied levels) and inverse photoemission (probing unoccupied levels) spectroscopy. The energies of the probes or signals involved make these methods particularly suited for surface analysis. In addition, if the photoelectron intensity in PES is also measured in an angle-resolved manner (ARPES), both its energy and momentum can be determined simultaneously, allowing the complete determination of the band structure of the solid. Alternatively, the unoccupied electronic states of the material are probed by XAS that follows dipole selection rules. The fine structure at the onset of the characteristic X-ray edge XANES provides information on the final density of unoccupied states, the transition probabilities, and many body effects. Further, the long-range oscillations extending from ∼50 eV for hundreds of eV (extending X-ray absorption fine structure, EXAFS) from the onset of the characteristic absorption edge, arise from interference effects of the wave scattered by the absorber atom with those back-scattered from neighboring ones. Thus, analyzing EXAFS data provides important element-specific structural information, e.g. nearest neighbor distances and their coordination number. A significant advantage of EXAFS over diffraction methods (§4, §7, and §8) is that it does not require a periodic crystal, and as such finds particular use in dynamic measurements of reactions and processes in structural biology, biochemistry, catalysis, and electrochemistry.

OUP CORRECTED – FINAL

Further Reading 211 .............................................................................. FURTHER READING

Bernath, P. F. Spectra of Atoms and Molecules. Oxford: Oxford University Press, 1995. Borg, R. J., and G. J. Dines. The Physical Chemistry of Solids. New York: Academic Press, 1992. Callister, W. D. Jr., and D. G. Rethwisch. Materials Science and Engineering. Hoboken: Wiley, 2010. Cox, P. A. The Electronic Structure and Chemistry of Solids. Oxford: Oxford University Press, 1987. Cothup, N. B., L. H. Daly, and S. E. Wiberley. Introduction to Infrared and Raman Spectroscopy. New York: Academic Press, 1990. Duckett, S., and B. C. Gilbert. Foundations of Spectroscopy. Oxford: Oxford University Press, 2000. Egerton, R. F. Electron Energy-Loss Spectroscopy in the Electron Microscope. Boston: Plenum Press, 1996. Ferraro, J. R., K. Nakamura, and C. W. Brown. Introductory Raman Spectroscopy. New York: Academic Press, 2003. Flewitt, P. E. J., and R. K. Wild. Physical Methods for Materials Characterization. Boca Raton: IOP, 2003. Fuggle, J. C., and J. E. Inglesfield, eds. Unoccupied Electronic States: Fundamentals for XANES, EELS, IPS and BIS. Berlin: Springer-Verlag, 1992. Harwood, L. M., and T. D. W. Claridge. Introduction to Organic Spectroscopy. Oxford: Oxford University Press, 2003. Hendra, P., C. Jones, and G. Warnes, Fourier Transform Raman Spectroscopy, New York: Ellis Horwood, 1991. Hüfner, S.Photoelectron Spectroscopy: Principles and Applications. Springer, 2003. Ibach, H., and H. Lüth. Solid-State Physics. Berlin: Springer, 1991. Kittel, C. Introduction to Solid-State Physics. Chichester: Wiley, 1986. Krishnan, K. M. Fundamentals and Applications of Magnetic Materials. Oxford: Oxford University Press, 2016. Leng, Y.Materials Characterization. New York: Wiley-VCH, 2013. Long, D. A.Raman Spectroscopy. New York: McGraw–Hill, 1977. McCreery, R. L. Raman Spectroscopy in Chemical Analysis. New York: Wiley, 2000. Papaconstantopoulos, D. A. Handbook of Band Structures of Elemental Solids: From Z = 1 To Z = 112, 2nd ed. New York: Springer, 1986. Pettifor, D. Bonding and Structure of Molecules and Solids. Oxford: Oxford University Press, 1995. Schlachter, A. S., and F. J. Wuilleumier, eds. New Directions with Third-Generation Soft X-Ray Synchrotron Radiation Sources. Dordrecht: Kluwer Academic Publishers, 1994.

OUP CORRECTED – FINAL

212 Bonding and Spectra of Molecules and Solids Shackelford, J. F. Introduction to Materials Science for Engineers. New York: Prentice Hall, 1996. Siegbahn, K.ESCA: Atomic, Molecular, and Solid-State Structure Studied by Means of Electron Spectroscopy. Uppsala: Almqvist & Wiksells, 1967. Stohr, J. “EXAFS and Surface EXAFS: Principles, Analysis, and Applications.” In Emission and Scattering Techniques, edited by P. Day. Dordrecht: D. Reidel Publishing, 1981. Sproul, R. L., and W. A. Philips. Modern Physics. Hoboken: Wiley, 1980. Sutton, A. P. Electronic Structure of Materials. Oxford: Oxford University Press, 1993. Teo, B. K., and D. C. Joy. EXAFS Spectroscopy. New York: Springer, 1981. Willard, H. H., L. L. Merritt Jr., J. A. Dean, and F. A. Settle Jr.Instrumental Methods of Analysis. Belmont: Wadsworth Publishing, 1988. Watts, J. F. An Introduction to Surface Analysis by Electron Spectroscopy. Royal Society Microscopy Handbook 32. Oxford: Oxford University Press, 1990. .............................................................................. REFERENCES

[1] Jensen, E., and E. W. Plummer. “Experimental Band Structure of Na.” Physical Review Letters 55 (1985): 1912. [2] Plummer, E. W. “Deficiencies in the Single-Particle Picture of Valence-Band Photoemission.” Surface Science 152/153 (1985): 162. [3] Himpsel, F. J. “Exchange Splitting of Epitaxial fcc Fe/Cu(100) versus bcc Fe/Ag(100).” Physical Review Letters 67 (1991): 2363. [4] Himpsel, F. J., E. Ortega, G. J. Mankey, and R. F. Willis. “Magnetic Nanostructures.” Advances In Physics 47 (1998): 511–97. [5] Schneider, C. M., and J. Kirschner. “Spin- and Angle-Resolved Photoelectron Spectroscopy from Solid Surfaces with Circularly Polarized Light.” Critical Reviews in Solid State and Materials Sciences 20, (1995): 179–283. [6] Fanconi, B. M. “Fourier Transform Infrared Spectroscopy of Polymers— Theory and Application.” Journal of Testing and Evaluation 12, no. 1 (1984): 33–9. [7] Brosseau, C. L., F. Casadio, and R. P. van Duyne. “Revealing the Invisible: Using Surface-Enhanced Raman Spectroscopy to Identify Minute Remnants of Color in Winslow Homer’s Colorless Skies.” Journal of Raman Spectroscopy 42, (2011): 1305–10. [8] Hüfner, S., and G. K. Wertheim. “X-Ray Photoemission Studies of the 3d Metals from Mn to Cu.” Physics Letters A 47, no. 5 (1974): 349–50.

OUP CORRECTED – FINAL

References 213 [9] Turtle, R. R., and R. J. Liefeld. “Densities of Unfilled One-Electron Levels in the Elements Vanadium and Iron through Zinc by Means of X-Ray Continuum Isochromats.” Physical Review B 7, (1973): 3411. [10] Yu, K. J., J. C. McMenahim, and W. E. Spicer. “UPS Measurements of Molecular Energy Level of Condensed Gases.” Surface Science 50 (1975): 149–56. [11] van der Laan, G., C. Westra, C. Haas, and G. A. Sawatzky. “Satellite Structure in Photoelectron and Auger Spectra of Copper Dihalides.” Physical Review B 23 (1981): 4369. [12] Olmstead, M. A., R. I. G. Uhrberg, R. D. Bringans, and R. Z. Bachrach. “Photoemission Study of Bonding at the CaF2 -on-Si(111) Interface.” Physical Review B 35 (1987): 7526. [13] Bianconi, A. “Core Excitons and Inner Well Resonances in Surface Soft X-Ray Absorption (SSXA) Spectra.” Surface Science 89 (1979): 41–50. [14] Pearson, D. H., C. C. Ahn, and B. Fultz. “White Lines and d-Electron Occupancies for the 3d and 4d Transition Metals.” Physical Review B 47, (1993): 8471. [15] Stohr, J. “Exploring the Microscopic Origin of Magnetic Anisotropies with X-Ray Magnetic Circular Dichroism (XMCD) Spectroscopy.” Journal of Magnetism and Magnetic Materials 200, (1999): 470–97. [16] Yano, J., and V. Yachandra. “X-Ray Absorption Spectroscopy.” Photosynthesis Research 102, no. 2–3 (2009): 241–54. [17] Murray, C. B., D. J. Norris, and M. G. Bawendi. “Synthesis and Characterization of Nearly Monodisperse CdE (E = S, Se, Te) Semiconductor Nanocrystallites.” Journal of the American Chemical Society 115 (1993): 8706–15. [18] Hüfner, S., S. Schmidt, and F. Reinert. “Photoelectron Spectroscopy—An Overview.” Nuclear Instruments and Methods in Physics Research A 547, no 1 (2005): 8–23. [19] Brille, E., and I. Arkin. “Site-Specific Hydrogen Exchange in a Membrane Environment Analyzed by Infrared Spectroscopy.” The Journal of Physical Chemistry Letters 9 (2018): 4059–65. [20] Nagai, Y., T. Hirabayashi, K. Dohmae, N. Takagi, T. Minami, H. Shinjoh, and S. Matsumoto. “Sintering Inhibition Mechanism of Platinum Supported on Ceria-Based Oxide and Pt-Oxide–Support Interaction.” Journal of Catalysis 242, no. 1 (2006): 103–9. [21] Gandubert, A. D., C. Legens, D. Guillaume, S. Rebours, and E. Payen. “Xray Photoelectron Spectroscopy Surface Quantification of Sulfided CoMoP Catalysts – Relation Between Activity and Promoted Sites – Part I: Influence of the Co/Mo Ratio.” Oil & Gas Science and Technology 62, no. 1 (2007): 79–89.

OUP CORRECTED – FINAL

214 Bonding and Spectra of Molecules and Solids .............................................................................. EXERCISES

A. Test Your Knowledge For each statement, identify ALL the correct possibilities. 1. Core and valence electrons (i) are the same; they are indistinguishable. (ii) are not affected by bonding. (iii) behave differently on bonding. Core electron energies are barely perturbed while valence electrons participate actively in the bonding. 2. For polyatomic molecules, the total number of molecular orbitals (i) depends on the type of bonding. (ii) depends on the temperature. (iii) is the same as the total number of valence orbitals of all the atoms involved in the bonding. 3. Energy bands in solids (i) are groups of electrons that make quantum music. (ii) have widths that depend on the strength of interactions between neighboring atoms. (iii) determine electronic properties of materials. 4. The density of states of a solid (i) determines whether it will float or sink in water. (ii) describes the distribution of electronic energy levels in the material. (iii) in the free electron model shows a E 1/2 dependence. 5. The three primary types of bonds (i) are ionic, covalent, and van der Waals. (ii) necessarily involve transfer or sharing of electrons. (iii) largely influence materials properties. 6. All bonding types (i) involve attractive and repulsive forces. (ii) have an attractive potential proportional to 1/R. (iii) consider only short-range interatomic interactions. 7. Covalent and ionic bonds (i) are both directional bonds. (ii) are determined by optimal space-filling considerations. (iii) differ in both directionality of the bonds and the importance of spacefilling considerations. 8. In molecules and solids (i) electronic, vibrational, and rotational energy levels are separated by the same magnitude.

OUP CORRECTED – FINAL

Exercises 215

9.

10.

11.

12.

13.

14.

15.

16.

(ii) the vibrational modes can be approximated by a classical harmonic oscillator model. (iii) even at T = 0 K, the molecules vibrate with the zero-point energy. Vibration modes in a molecule (i) are related to the number of degrees of freedom. (ii) involve stretching, bending, rocking, and scissoring. (iii) are sensitive to temperature. Molecular vibrations (i) lead to Raman and infrared activity in ALL molecules. (ii) with a derivative of the dipole moment lead to Rayleigh scattering. (iii) with a derivative of the dipole moment lead to IR activity. (iv) with a derivative of the polarization lead to Stokes scattering. (v) with a derivative of the polarization lead to Raman scattering. (vi) give the same intensity for both Stokes and anti-Stokes scattering. Using a Michelson interferometer in FTIR allows (i) signal modulation and a measurable frequency consistent with detector response. (ii) higher signal collection speeds with improved signal-to-noise ratio. (iii) allows for setting up the experiment inexpensively. Resonance enhancement in Raman scattering (i) increases the signal by orders of magnitude. (ii) often produces additional bands in the signal. (iii) in practice is affected by competing fluorescence effects. In practical implementation of Raman spectroscopy (i) doubling the focal length of the grating doubles the spectral resolution (ii) lasers in the visible spectrum are avoided for safety reasons. (iii) Rayleigh line is NOT a problem, and can be ignored. (iv) the surface on which the molecules are adsorbed does not influence the signal. The band gap of semiconductors can be determined by (i) optical spectroscopy. (ii) PES. (iii) XAS. Important energy levels in a solid for spectroscopy are (i) vacuum and Fermi levels INSIDE the solid. (ii) vacuum and Fermi levels OUTSIDE the solid. (iii) the Fermi level inside and the vacuum level outside the solid. The work-function (i) defines the work involved in doing any spectroscopy measurement. (ii) is the difference in energy between the vacuum and Fermi levels of a solid. (iii) is of the order of 2–5 eV in most metals. (iv) is of no practical importance, especially in electron guns.

OUP CORRECTED – FINAL

216 Bonding and Spectra of Molecules and Solids 17. It is best to use (i) PES to probe filled levels in the valance bands. (ii) inverse PES to probe unfilled levels in the valence band. (iii) XAS to probe unoccupied levels. 18. The convention for binding energy, EB , is (i) arbitrary. (ii) such that EB = 0, at E = EF in solids. (iii) such that EB = 0, at E = EV in molecules. (iv) such that EB > 0, for PES. 19. In PES, it is best to (i) use UV light to investigate core levels. (ii) use UV light to investigate the valence band. (iii) use UV light only if an appropriate X-ray source is not available. (iv) set the onset of the signal to be at E = EF . (v) assume that there is no background from secondaries. (vi) use a shallow angle of incidence to enhance surface sensitivity. (vii) ignore the effects of ligands in interpreting the spectrum. (viii) interpret the data as sensitive only to the surface. 20. Inverse photoemission is (i) implemented in only one well-defined way. (ii) best used to study occupied states with E < EF . (iii) best used to study unoccupied states with EF < E < EV . (iv) is complementary to PES. (v) probes only those levels allowed by dipole selection rules. 21. The Fermi–Dirac distribution (i) can be used to measure the resolution of a PES spectrometer. (ii) is only of academic interest to physicists. (iii) gives the probability of the occupancy of electronic levels as a function of T. 22. In XAS (i) dipole selections determine allowed transitions. (ii) data interpretation requires knowledge of BOTH initial and final states. (iii) data interpretation can be based on final unoccupied DOS. 23. XAS (i) is exactly the same as inverse PES. (ii) is similar to electron energy-loss spectroscopy. (iii) often requires a synchrotron radiation source. (iv) can only be measured in transmission through a very thin specimen. (v) has no depth sensitivity and is a surface analysis technique. 24. The fine structures in XAS (i) are of two kinds: near-edge and extended. (ii) cannot give any information on the bonding.

OUP CORRECTED – FINAL

Exercises 217 (iii) can give information on the nearest neighbor distances and coordination numbers.

B. Problems 1. Classic harmonic oscillator: for the harmonic oscillator describing the vibrations of two atoms, what will happen to the vibration frequency (a) if one of the atoms is replaced with one of heavier mass? (b) if the bonding becomes weaker, but keeping the masses of the two atoms the same? In addition, what is the reduced mass for (c) a homonuclear diatomic molecule, with atomic mass, m? (d) the molecule HX, where mX >> 1? 2. The potassium hydride molecule, K–H, has an equilibrium bond length of 2.32 Å. It undergoes a rotational transition from J = 1 to J = 2. What is the frequency of the photon that it needs to absorb to make this transition possible? 3. X-ray sources: The 2p peak in XPS for a given element using Al Kα (1487 eV) is at a kinetic energy of 700 eV. Instead, if Mg Kα (1,254 eV) were to be used, where will the same peak show up in the spectrum? 4. The photoelectron energy spectrum of an unknown specimen, using Mg Kα irradiation at 1.250 keV is shown in Figure 3.Pr.4. The energy axis is EB = hf —Ekin , and the vertical axis is the observed intensity in arbitrary units. Two regions of the spectrum have been enhanced for clarity.

1.25 keV Radiation (Mg Kα) Intensity (a.u)

17.4 878

858 868 848 Binding energy, eV

×4

c 1100 1000

900

800

700 600 500 400 Binding energy (eV)

300

200

100

0

OUP CORRECTED – FINAL

218 Bonding and Spectra of Molecules and Solids (a) Index all the major features indicated by arrows. (b) Now, identify the material. 5. FTIR with a Michelson interferometer: your FTIR detector electronics can respond up to a maximum frequency of 1 kHz. The mirror in the interferometer can be moved within a velocity range of 0.1–0.75 cm/s. (a) What is the range of incident photon wavelengths that can be used in this set-up? (b) Will this instrument be capable of covering the mid-IR from the group-frequency to thefinger-printing regime? 6. The X-ray absorption spectrum (actually, this is EELS data, but for your purposes they are the same) of three forms of carbon (amorphous, graphite, and diamond) are shown in Figure 3.Pr.6.

X-ray absorption (a.u.)

Diamond

Graphite

Amorphous Carbon 290 300 310 320 Photon energy (eV) (a) Identify the edge (K, L, etc.). (b) Draw a simple molecular orbital (MO) energy diagram of sp2 and sp3 hybridized carbon in graphite and diamond, respectively. (c) See if you can associate the main features in the absorption spectrum with the levels in your MO picture. 7. Using the plot (Fig. 3.Pr.7a) that gives various possible characteristic bands in vibrational spectra for diatomic stretching vibrations, see if you can identify any of the features in the IR and Raman spectra of (a) CHCl3 in Figure 3.4.11 (b) styrene/butadiene rubber shown in Figure 3.Pr.7b. If not, look up any reference “finger print” data, e.g. Colthup et al. (1990), and see if you can do better.

OUP CORRECTED – FINAL

Exercises 219 C–D

C–H

C–F

Single bond C–H

C–CI C–Br C–I C–O C–C C–N

X–H

O–H

P–H S–H

Double or trible x=y or x≡y

N–H

C≡C

C=C

C≡N

C=N C=O

4000

C–S

S–H

3000

C=S P=O S=O

2000 cm–1

1000

0

Infrared

Raman

3200

2800

2400

2000

1600 cm–1

1200

800 600

300

8. X-ray photoemission spectroscopy using Al Kα radiation. In your spectrum you see peaks at 942 eV, 778 eV, 765 eV, and 643 eV. What are the binding energies of these electrons? Identify the two elements present in your specimen. Hint: Use Table 2.2.2. 9. Michelson interferometer: An He–Ne laser based FTIR detector has a mirror that can be moved with a velocity in the range of 0.2–0.8 cm/s. What maximum response frequency is required for this device if it has to detect wavenumbers in the range 400–4,000 cm–1 ? Will this device cover the mid-IR region, from the group frequency to the finger-printing regime? 10. Can IR spectroscopy detect a transition with an energy of 325 kJ/mol?

OUP CORRECTED – FINAL

Crystallography and Diffraction

4 4.1 The Crystalline State

222

4.2 The Reciprocal Lattice

250

4.3 Diffraction

256

4.4 Quasicrystals and the Definition of a Crystalline Material

266

Summary

267

Further Reading

268

References

269

Exercises

270

The structure and diffraction pattern of a spinel (MgAl2 O4 ) crystal. Top: Drawing of the cubic unit cell, in real space, where red spheres represent the oxygen atoms, Mg atoms sit within the yellow tetrahedral sites, and Al atoms sit within the blue octahedral sites. A unit cell contains eight formula units. The space group of this crystal is Fd3m. Bottom: A transmission electron diffraction pattern, in reciprocal space, obtained with a convergent beam (CBED) incident along the [001] direction of the spinel crystal. Such CBED patterns are discussed further in §8.6.3.

Principles of Materials Characterization and Metrology. Kannan M. Krishnan, Oxford University Press (2021). © Kannan M. Krishnan. DOI: 10.1093/oso/9780198830252.003.0004

OUP CORRECTED – FINAL

Crystallography and Diffraction 221 Chapter 3 introduced different types of bonding in materials and developed spectroscopic methods to analyze them. We saw that the type of bonding determines the local arrangements of atoms including their coordination number and interatomic distances. The bonding also determines the magnitude of a given property of the material, and the local symmetry or point group. The latter, in turn, defines the components of the tensor that relates the response of the materials to an applied stimulus (see Nye, 2000). An example is the second-rank, polarizability tensor, αij , relating the applied field, E, and the dipole moment, μ, introduced in §3.4.2. In some materials, the arrangement of the atoms determined by the bonding is limited to their immediate environment; materials with such short-range order are commonly referred to as amorphous, or formless, materials. In many other materials, the local atomic arrangement is spatially repetitive in a periodic way, and extends over the macroscopic extent of the crystal. Materials with such longrange order are traditionally called crystalline materials, but recent discovery of quasicrystalline materials (see §4.4) has led to a redefinition of what makes a material crystalline. Nevertheless, it is important to be able to rigorously describe the entire class of periodic crystalline materials, i.e. be familiar with the field of classical crystallography, as it is the foundation of structural characterization of materials by diffraction. Note that, even though it may initially appear that a very large number of periodic crystal structures are to be found in nature or synthesized as engineering materials, they can be classified in three dimensions into a small number of groups based on local, or point group symmetry (32 in number), and the patterns, or space groups (230 in total), they generate by periodic repetition. Crystallography is a highly developed subject that is discussed in detail in a number of excellent books; see, among others, Allen and Thomas (1998), Borchard-Ott (1995), Buerger (1970), Hammond (2006), and Sands (1975). In the context of materials characterization and analysis, our interest and scope here is modest. We only wish to use crystallography as a language to describe crystalline materials in the context of developing diffraction/scattering methods to elucidate their structure and symmetry in appropriate detail. In this chapter, after an overview of periodic crystal structures, we discuss the important concept of a reciprocal lattice. Then, we introduce the basic principles of diffraction in both real and reciprocal space and discuss it qualitatively for both X-ray and electron incidence. We conclude this chapter with a very brief introduction to an interesting family of materials, called quasicrystals, discovered by Professor Dan Schechtman.1 Detailed quantitative treatment of diffraction, and different scattering geometries follows for X-rays in §7, and electrons and neutrons in §8.

1 Dan Shechtman (1941–) won the Nobel prize in Chemistry in 2011, and is cited for “the discovery of quasicrystals.”

OUP CORRECTED – FINAL

222

Crystallography and Diffraction

4.1 The Crystalline State We begin with some simple definitions. A crystal is a homogeneous body composed of a periodic arrangement of atoms, ions, or molecules, in three dimensions with anisotropic properties. By anisotropic properties we mean that the properties of a crystal are different when measured along different directions. However, this is by no means universal, as there also exist isotropic crystals, where the physical property is the same when measured in any direction. If the periodic arrangement in a solid is perfect, and extends uniformly throughout the entire sample in all directions, we refer to it as a single crystal. Such single crystals are common in nature, and high quality, defect-free ones are also synthesized artificially in the laboratory. In materials research we also encounter them as thin films grown epitaxially on other single crystal substrates. However, most engineered crystalline solids used in practice are made up of small units of single crystals, or grains, on the micrometer length scale. These materials are termed polycrystalline, and they are characterized by random crystallographic orientations; the regions where different grains meet in a solid are called grain boundaries (Fig. 4.1.1; §4.1.9.3). In some cases, the grains in a polycrystalline material exhibit a preference for a particular crystallographic orientation (these

Figure 4.1.1 Schematic representation of the stages in the solidification of a polycrystalline material; each square grid represents a unit cell: (a) Small nuclei. (b) Early stages of growth, with obstruction of some grains. (c) After solidification with the formation of irregular-shaped grains; the size of the grains determines the extent of long-range order, and also affects the full-width at half maximum (FWHM) of diffraction peak intensities. (d) Grain boundaries of (c) visible in an optical microscope. Further, the composition and structure of individual grains and grain boundaries can be characterized, at atomic resolution, by various transmission electrons microscopy (§9) methods. Adapted from Callister and Rethwisch (2009).

(a)

(b)

(c)

(d)

OUP CORRECTED – FINAL

The Crystalline State 223

(a)

(b)

Silicon atom Oxygen atom

Figure 4.1.2 A schematic representation of SiO2 in (a) crystalline form, and (b) non-crystalline form. Note that the nearest neighbor coordination in both cases are the same. Adapted from Callister and Rethwisch (2009).

terms will be defined in §4.2), and such materials are said to have a texture. As we will see, diffraction methods probe the long-range order (LRO) in the solid, and the grain size, which defines the spatial extent of the long-range order, will determine the broadening (FWHM) of the diffraction peaks (§7.8). If the temperature is low enough, and given sufficient time, all materials will eventually crystallize, since the ordered crystalline state is the one with the lowest energy. However, some solids, such as glass, never reach the lowest energy state; thus, they have higher energy than the equivalent crystalline state and are referred to as noncrystalline or amorphous materials. In such amorphous materials, the building blocks defined by the bonding are retained (Fig. 4.1.2) and the shortrange order (SRO) is preserved, but the crystallinity or long-range order is not developed. Such SRO (§7.11.5) has its own signature in the form of diffuse intensities in diffraction patterns (Warren, 1990).

4.1.1 Lattices Regularity and repetition are the central features of a periodic crystal structure. To appreciate these factors, the atoms in the crystal can be replaced by points, such that each point in space has a fixed relationship to the atoms in the crystal. In this manner, we can generate a space lattice, defined as an array of points arranged in space with identical surroundings; Figure 4.1.3a shows it in its most general form. This lattice can also be visualized as a repeating set of unit cells, in the form of parallelepipeds (Fig. 4.1.3b), which are identical in size, shape, and orientation with respect to their neighbors. Now, there are two ways to describe the size and shape of the unit cell. One way is to describe it using a set of three, non-coplanar vectors (a, b, and c) drawn from one corner of the unit cell and defined as the crystallographic axes of the cell. Further application of the lattice translation to these vectors, i.e. Pa + Qb + Rc, where P, Q, and R, are integers, will generate all the points of the space lattice. Alternatively, the repeating unit cell geometry is completely defined in terms of six parameters, i.e. the three edge lengths (a, b, and c), and the three inter-axial angles (α, β, and γ ) between them (Fig. 4.1.3b).

OUP CORRECTED – FINAL

224

Crystallography and Diffraction

Figure 4.1.3 (a) A space lattice with one unit-cell highlighted (b) The highlighted unit cell with all the six parameters indicated. Note: a two-dimensional or surface lattice can be defined by only three variables, a, b, and γ . See Figure 4.1.15 and §8.3.1. Adapted from Cullity (1978).

(a)

(b) c c E

β α a γ

b

b

a

Note that every unit cell has eight identical vertices and six faces. However, a lattice point at any one of the vertices (say E, in Fig. 4.1.3a), is shared by eight unit-cells that meet at that point. Thus, only one eighth of the lattice point at any of the eight vertices of the parallelepiped may contribute to a specific unit cell. Since, there are eight vertices or corners, N C , each contributing one eighth, in effect the unit cell contains a single lattice point and is defined as a single or primitive unit cell, with the crystallographic symbol, P. Now, we generalize this concept to classify all crystals in three dimensions into fourteen Bravais lattices and seven crystal systems.

4.1.2 Generalized Crystal Systems and Bravais Lattices

2 In any system, if a = b = c, then they are all given the same symbol, a, to indicate the equality. 3 A. Bravais (1811–1863) was a French mathematician and crystallographer.

The unit cell geometry, completely defined by the six parameters (a, b, and c, and α, β, and γ ) in Figure 4.1.3, gives rise to seven possible combinations, each of which represents a distinctly different crystal system. These crystal systems, in decreasing order of symmetry, where cubic2 (a = b = c and α = β = γ = 90◦ ) is the highest, and triclinic (a  = b = c, α = β = γ ) is the lowest, are listed in Table 4.1.1. Note that the rhombohedral (R) system, can be considered as a subset of the hexagonal system; in this case, the number of crystal systems reduces from seven to six. By periodically repeating the unit cells of the seven crystal systems, we can generate seven different primitive lattices (if rhombohedral (R), is excluded, then we have six primitive lattices). However, as defined, the primary requirement of a space lattice is that each point has identical surroundings. Now, additional variations (nonprimitive) of these seven lattices can be generated, but it turns out that there are only seven more nonprimitive possibilities. Thus, in total we have fourteen Bravais3 lattices (Fig. 4.1.4). In addition to the simple or primitive (P) lattice, already discussed in the last section, additional face centered (F), body centered (I), and base centered (A, B, or C) lattices are introduced. The F-lattice includes additional lattice points, NF , occupying face centers and shared by two unit-cells, and the I-lattice includes an additional lattice point, NI , at the exact center of the unit cell that is not shared with any other. In general, the total number of lattice point, N, per unit cell is given by

OUP CORRECTED – FINAL

The Crystalline State 225 Table 4.1.1 Crystal systems (7) and Bravais lattices (14), their characteristic symmetries, and point groups (32), in threedimensions. System

Bravais lattice

Axial lengths and angles

Characteristic (minimum) symmetry

Noncentrosymmetric Centrosymmetric point groups point groups

Cubic

Simple, P Body centered, I Face centered, F

a=b=c α = β = γ = 90◦

23 432 43m

m3 m3m

Tetragonal

Simple, P Body centered, I

a = b = c α = β = γ = 90◦

Simple, P Body centered, I Face centered, F Base centered, C

a = b = c α = β = γ = 90◦

4 422 4 4mm, 42m 222 mm2

4/m 4/mmm

Orthorhombic

4 threefold rotation (triad) axes at 109.47◦ 1 fourfold rotation* (tetrad) axes 3 twofold rotation* (diad) axes at 90◦ (set of 3 orthogonal mirrors) 1 threefold rotation* (triad) axis 1 sixfold rotation* (hexad) axis

Rhombohedral Simple (Trigonal)

a=b=c α = β = γ  = 90◦

Hexagonal

Simple

a = b = c α = β = 90◦ , γ = 120◦

Monoclinic

Simple, P Base centered, C

a = b = c α = γ = 90◦  = β ≥ 90◦

Triclinic

Simple

a = b = c α = β = γ = 90◦

* with or without inversion Adapted from Hammond (2006).

1 twofold rotation* (diad) axes None

mmm

3 32 3m

3 3m

6 622 6 6mm 6m2 2 m

6/m 6/mmm

1

1

2/m

OUP CORRECTED – FINAL

226

Crystallography and Diffraction

a

a

a a

a a

a

a

a

Body-centered cubic (I )

Simple cubic (P)

Face-centered cubic (F )

c c

c a

c b

a

a

b a

a

Simple tetragonal (P)

Body-centered tetragonal (I )

c

c b

Body-centered orthorhombic (I )

a a αa α α

120° a

c

a Face-centered orthorhombic (F)

Base-centered orthorhombic (C)

a

Hexagonal (P)

Rhombohedral (R)

c

c

Figure 4.1.4 The seven crystal structures that give rise to the fourteen Bravais lattices as presented in Table 4.1.1.

Simple orthorhombic (P)

b

a

β

a

b

a Simple monoclinic (P)

c β

b

a Base-centered monoclinic (C) N = NC /8 + NF /2 + NI

β

α

b

a γ Triclinic (P)

(4.1.1)

where NC , as mentioned, is the total number of points on the corners of the unit cell. In addition to I and F, possible nonprimitive arrangements (Fig. 4.1.4; Table 4.1.1) are the A, B, and C type, base centered lattices, where additional

OUP CORRECTED – FINAL

The Crystalline State 227 lattice points are centered on any one pair of faces (note, the A face is defined by the b and c axis, and the others are defined cyclically). Needless to say, the periodic application of the unit-cell vectors (a, b, and c) to all points, wherever they are located in the cell, can also extend to all the nonprimitive unit cells through space. Real crystal structures are generated from the Bravais lattices by having atoms, ions, or molecules, occupy all the points of the space lattice. The arrangement of atoms associated with every lattice point is called a basis. Then, by applying the lattice translations that replicate the atoms throughout the lattice, a crystal structure is generated. In other words, we have Lattice + basis = crystal

c

(4.1.2)

We should also point out that any of the non-primitive Bravais lattices can be referred to an equivalent primitive unit cell; Figure 4.1.5 shows an example for the face centered cubic (FCC) structure. This will be important later in §7 and §8 as we “index” diffraction patterns of materials.

b 60° O

a

Figure 4.1.5 The face centered cubic (FCC) lattice referred to cubic (multiple) and rhombohedral (unit) cells.

4.1.3 Lattice Points, Lines, Directions, and Planes Every lattice point in the crystal can be uniquely defined, with respect to the origin of the space lattice, by the vector, τ = ua + vb + wc, where a, b, and c, are unit cell vectors with lengths, a, b, and c (Fig. 4.1.6). Thus, only the coordinates, u v w, are required to specify a point; if u, v, and w, are integers, they represent points on the primitive or P-lattice. Alternatively, u,v,w can also have fractional values of ½, ¼, 1/3, or 2/3; in such cases, they represent points within the unit cell. Mathematically, in any coordinate system, two points can specify a line. In crystallography, the direction of any line in a lattice is described by drawing a line parallel to it but passing through the origin, and then using the coordinates of any lattice point on the line. Thus, if the line passes through 000 and uvw, where the latter need not be integers, then [uvw] are indices of the direction of the line

011

001 101 c a 100

000

111 010

b 110

τ

231

Figure 4.1.6 Lattice points are defined by the vector from the origin to the point, uvw. Here, τ defines the lattice point 231. Adapted from Borchardt–Ott (1995).

OUP CORRECTED – FINAL

228

Crystallography and Diffraction – [100]

– [233]

– –23 11

[111]

[001] [100] [010]

c

[120] [210]

b Figure 4.1.7 Indices, [uvw], for different directions. Adapted from Cullity (1978).

1 –12 – a [120]

0

[100]

(Fig. 4.1.7). In practice, the values of [uvw] are converted to a set of smallest integers; hence, [1/2 ¼ 1], [214], and [1½2], all refer to the same line, but [214], with all integers, is preferred. Negative indices are indicated with a line or bar on top, i.e. [uvw]. Many lines are generated by the symmetry of the crystal, and those ones that are related by symmetry are represented by angular brackets; hence, in a cubic system, represents the family of six face diagonals—[110], [101], [011], [110], [101], [011]. The orientation of a plane in a lattice can be represented symbolically as (hkl), based on its intercepts with the three principal axes. Since a plane parallel to a specific crystallographic axis will only “intercept” it at infinity, the convention is to use the reciprocal of its intercept. Thus, we define the Miller4 indices, (hkl), for a lattice plane as the reciprocal of its intercept with the three crystallographic axes. Again, in practice, the convention is to choose (hkl) such that they are the smallest integral multiples of the reciprocals of the plane intercepts (Fig. 4.1.8). Example 4.1.1: Calculate the Miller indices of the planes shown in Figure 4.1.8a. Solution: The intercepts, PQR, of the planes are 111 (A), and 2 2.5 3.5 (B) The equation of the plane is given by

x P

+

y Q

+

z R

= 1.

Multiplying by PQR, we get xQR + yPR + zPQ = PQR = xh + yk + zl. Substituting the values of P, Q, and R, we get

4 W. H. Miller (1801–1880), who first used this notation, was an English mineralogist and crystallographer.

for Plane A: x + y + z = 1, which gives its Miller index as h = k = l = 1, or (111) y z for Plane B: 2x + 2.5 + 3.5 = 1, or 8.75x + 7 y + 5 z = 17.5, and multiplying by 4 to set all coefficients as the smallest integers, we get 35x + 28y + 20z = 70, giving the Miller index of (35 28 20).

OUP CORRECTED – FINAL

The Crystalline State 229 c (a)

z

b

(b) d100

d200

(100)

(200)

– (110)

– (111)

a

(110)

x y

(102)

Figure 4.1.8 Miller indices of planes. (a) Two planes with intercepts of 111 and 2 2.5 3.5 are shown. Their indices are (111) and (35 28 20). (b) Miller indices of some lattice planes and the inter-planar spacing, dhkl . See Table 4.1.2.

In reality, for any plane in any lattice, there must exist a whole set of parallel planes that are equidistant, one of which passes through the origin. Also, planes with Miller indices (nh nk nl) are parallel to (hkl), but with 1/nth the inter-planar spacing. The inter-planar spacing, dhkl , in crystals is important in diffraction, as we will see in §4.2, §7, and §8. They are tabulated, in terms of the unit cell parameters, for all seven crystal systems in Table 4.1.2. Finally, it must be mentioned that only in cubic systems, the direction [hkl] is always normal to the plane (hkl). For the hexagonal system, consistent with its symmetry, planes can also be indexed by a different system, the Miller–Bravais indices, (hkil), using the inverse intercepts on all the four crystallographic axes, a1 , a2 , a3 , and c (Fig. 4.1.9). However, since a1 , a2 , and a3 , are coplanar and separated by 120◦ , it is easy to see that a1 + a2 = −a3 , and h + k = −i. Hence, the index, i, is redundant and is replaced by a dot, and the index of the plane is written as (hk.l); often, for further simplicity, the dot is not even used! As for the directions, [UVW], in the hexagonal system, they are expressed in terms of the three axes, a1 , a2 , and c. They can be inter-related to the indices [uvtw], in the four axes, a1 , a2 , a3 , and c system, by the simple relationships U=u−t V=v−t W=w

u = (2U − V)/3 v = (2V − U)/3 t = − (u + v) = − (U + V)/3 w=W

(4.1.3)

OUP CORRECTED – FINAL

230

Crystallography and Diffraction

Table 4.1.2 Interplanar spacing, dhkl , in terms of unit cell parameters for all crystals systems. Interplanar spacing, dhkl

Crystal system and their unit cell parameters Cubic Tetragonal Orthorhombic Hexagonal Rhombohedral Monoclinic Triclinic

a=b=c α = β = γ = 90◦ a = b = c α = β = γ = 90◦ a = b = c α = β = γ = 90◦ a = b = c α = β = 90◦ ; γ = 120◦

1 d2

=

h2 +k2 +l 2 a2

1 d2

=

h2 +k2 a2

+

1 d2

=

h2 a2

k2 b2

1 d2

=

4 3a2

a=b=c α = β = γ < 120◦  = 90◦ a = b = c α = γ = 90◦  = β a = b = c α = β = γ

1 d2

=

   2 2 2   2α 1 (1+cos α) h +k +l − 1−tan 2 (hk+kl+lh) a2 1+cos α−2cos2 α

1 d2

=

h2 a2 sin2 β

(a)

+

l2 c2

+

l2 c2



 h2 + hk + k2 +

+

k2 b2

+

l2 c2 sin2 β

l2 c2

−

2hl cos β ac sin2 β

  = V12 s11 h2 + s22 k2 + s33 l 2 + 2s12 hk + 2s23 kl + 2s31 lh where   V 2 = a2 b2 c2 1 − cos2 α − cos2 β − cos2 γ + 2 cos α cos β cos γ and s11 = b2 c2 sin2 α, s22 = c2 a2 sin2 β, s33 = a2 b2 sin2 γ , s12 = abc2 (cos α cos β − cos γ ) , s23 = a2 bc (cos β cos γ − cos α) s31 = ab2 c (cos γ cos α − cos β) 1 d2

z

(b)

z

(c)

z

[0001]

c

– (1010) a2

– [1120]

a3

a3 120°

– (1011)

a2

a1

– [1100]

a1

a3 (0001)

a1

Figure 4.1.9 Hexagonal crystals system. (a) Unit cell with the principal axes. Miller–Bravais indices for select (a) directions and (b) planes. Adapted from Callister and Rethwisch (2009).

OUP CORRECTED – FINAL

The Crystalline State 231 Example 4.1.2: Convert the following directions in a hexagonal crystal from the three-index to a four-index notation, applying (4.1.3): [123], [122], and [310]. Solution: (a) [123] = [UVW], thus u = 0, v = 1, t = −1, w = 3; thus   [uvtw] = 0113   (b) 122 = [UVW], thus u = 4/3, v = −5/3, t = 1/3, w = 2; thus  [uvtw] = 4516] (c) [310] = [UVW], thus u = 5/3, v = −1/3, t = −4/3, w = 0; thus  [uvtw] = 5140]

4.1.4 Zonal Equations In general, the equation of any plane may be written as x y z + + =1 m n p

(4.1.4)

where,xyz are the coordinates of any point lying on the plane, and m, n, and p, are the intercepts of the plane with the principal crystallographic axes. We have already defined (hkl) as the reciprocal of the intercepts, and so we can rewrite (4.1.4), in its general form as, hx + ky + lz = C

(4.1.5)

where C is an integer. Now, (4.1.5) represents a family of parallel planes, and for all h, k, l > 0, C = 1 defines the plane nearest the origin in the positive direction of a, b, and c. Similarly, C = −1, defines the plane nearest the origin, but in the negative a, b, and c directions. When C = 0, it defines the plane in the same family that passes through the origin and has the equation hx + ky + lz = 0

(4.1.6)

If any specific point, xyz, on the plane is defined by the lattice line, [uvw], we rewrite (4.1.6) as hu + kv + lw = 0

(4.1.7)

and refer it to as the zonal equation. Any two non-parallel planes, (h1 k1 l1 ) and (h2 k2 l2 ), will have a common intersection along a lattice line [uvw], given by the solution of h1 u + k1 v + l1 w = 0, and h2 u + k2 v + l2 w = 0, which is the cross product u = k1 l2 − k2 l1 v = l1 h2 − l2 h1 w = h1 k2 − h2 k1

(4.1.8)

OUP CORRECTED – FINAL

232

Crystallography and Diffraction [uvw]

[011] –– 111

– – 111

– 022

– 022 – 111

– 111

Figure 4.1.10 A common line, or zone axis, [uvw], at the intersection of different planes (only three planes are shown). On the right is an actual electron diffraction pattern from a single crystal of silicon. In this case, each diffraction spot represents a family of planes, (hkl), as indexed. All of them share a common line, [uvw] = [011], referred to as the zone axis that, here, is also the incident electron beam direction. Note that each of the planes satisfies the condition, uh + vk + wl = 0. See §8.6 for further details on transmission electron diffraction. The line [uvw] is referred to as the zone axis of (h1 k1 l1 ) and (h2 k2 l2 ). It is easy to see that a set of planes, consistent with the Bravais lattice, or the symmetry of the crystal, can be part of the same zone axis, provided the planes satisfy the condition (4.1.6). The concept of zone axis is of practical use in electron diffraction (Fig. 4.1.10; Fig. 8.6.3). Example 4.1.3: The following figure is a projection of a lattice along the c-axis, onto the a-b plane. The red lines labeled (A) and (B) are the traces of the planes parallel to the c-axis. (a) Index planes (A) and (B). (b) Calculate [uvw], the line common to the planes (A) and (B). (c) Draw the trace of the planes (320) and (120 on the same projection.

– (120)

(A)

O b a (320)

(B)

OUP CORRECTED – FINAL

The Crystalline State 233

Solution: The intercepts of the two planes are (A) 1 3 ∞, and (B) –1 2 ∞. Their indices, from (4.1.4) are (A) (310), and (B)(210). The common line is along the z-axis (normal to the plane shown). For (320) the intercepts are 2 3 ∞, and for (120), the intercepts are 2 –1 ∞. These are also plotted in the figure.

4.1.5 Atomic Size, Coordination, and Close Packing As shown previously, the structure of actual physical crystals can be related to the space lattice by placing basis atoms either at, or in some specific relation to, the points of the Bravais lattice. The process of combining a lattice and basis (4.1.2) generates a periodic arrangement of atoms in three dimensions to form a crystal. Representative examples of such structures are presented in §4.1.6. Here, we briefly introduce the additional concept of close packing in crystal structures, where the atoms/ions are considered as hard spheres that pack together to satisfy three important criteria: (i) fill space most efficiently, (ii) achieve an environment of the highest symmetry, and (iii) achieve maximum interaction with the largest coordination number of nearest neighbors. Note that the representation of atoms as hard spheres is not intended to show anything specific about their physical or chemical nature. Their diameters merely represent how closely they approach their neighbors, and depend on how they are packed together in a solid. Thus, the hard sphere model is not meant to be a representation of the structure of atoms or ions, but instead, a way to interpret their structure, mainly size, when they are packed together in a solid. In two dimensions, atoms can be arranged in a close-packed hexagonal or honeycomb pattern (Fig. 4.1.11a), which is the most compact way possible. We refer to this as the A layer. Note that there are two possible interstices or troughs, marked as B and C, where a second layer can be stacked. Without loss of generality, the second layer is stacked such that atoms are placed over the interstice B, and each atom touches three atoms in the layer below (Fig. 4.1.11b). For the third layer, there are two choices. They can sit in the interstices directly above the atoms in the first layer (Fig. 4.1.11c) giving the sequence ABABAB . . . , and form a hexagonal close packed (HCP) structure, with the unit cell shown in Figure 4.1.11e. Common metals with the HCP structure include Be, Mg, α–Ti, Co, and Zn. Alternatively, the third layer may not sit above the first layer, but instead sits over the interstices marked as C (Fig. 4.1.11d) giving a repeat sequence ABCABCABC . . . , and form a face centered cubic (FCC) structure, (also, known as cubic close-packed) with the unit cell shown in Figure 4.1.11f. FCC metals include Al and its alloys, γ –Fe, Ni, Cu, Pb, Ag, Au, Pt, etc. Note that both HCP and FCC structures are close-packed and equally dense.

OUP CORRECTED – FINAL

234

Crystallography and Diffraction ABABABAB............

(a)

(c)

H

(e)

F

G E

J

c D B

C A

(b)

(d)

ABCABCABCABC............

a

(f)

r

a a a

(g)

(h)

(i) a r

a a

Figure 4.1.11 (a) A single layer of a hexagonal close-packed plane of atoms, A; on the surface of this plane there are two sets, B and C, of equivalent troughs of low energy where the next layer of atoms can be placed. (b) The next layer of atoms placed in position, B. (c) The third layer can be placed on top of the first layer, A, and this sequence can be repeated to generate an ABAB stacking and a hexagonal close packed, HCP, structure. (d) Alternatively, the third layer can be placed at position, C, and the sequence repeated, to give an ABCABC . . . stacking and a face centered cubic, FCC, close packed structure.√(e) The unit cell of the HCP structure. (f ) The unit cell of the FCC structure. Notice that the face diagonal, 4r = 2a, where r is the radius of the atom. (g) Instead of the close-packed hexagonal array we can start with a less efficient, in terms of space-filling, square array of atoms. A second array can be placed exactly on top of this layer to form a simple cubic crystal (not shown). (h) A second plane can be placed in the positions of the troughs shown in (g), and the third layer can be placed back in position A. This gives the body √ centered cubic, BCC, structure. (i) The unit cell of the BCC structure. Note that now the body diagonal is 4r = 3a. Alternatively, in two dimensions, the first layer can form a square lattice arrangement (Fig. 4.1.11g) where there are a large number of interstices (marked as X). Now, the second layer can be stacked such that atoms are on top of each other and form a simple cubic structure (not shown in the figure). The other possibility is for the second layer to sit on the array of interstices (Fig. 4.1.11h) to form a body centered cubic (BCC) structure, with unit cell shown in Figure 4.1.11i. Note that neither the simple cubic, nor the body centered cubic, structure are close packed. However, a number of metals, notably Cr, Mo, α–Fe, and W, are found in the BCC structure.

OUP CORRECTED – FINAL

The Crystalline State 235 (a)

(c)

a√3/2 a/√2

a/2 a/2

a/√2

Metal atoms Tetrahedral sites Octahedral sites (b) a√3/4

a√5/4 a√3/2

a

(d) a/√2

Figure 4.1.12 Unit cell of the FCC structure showing (a) octahedral sites with rX /rA = 0.414, and (b) tetrahedral sites with rX /rA = 0.225. Similarly, for the BCC structure, (c) octahedral sites with rX /rA = 0.154, and (d) tetrahedral sites with rX /rA = 0.292, are shown. In general, in ionic compounds, we expect threefold, trigonal planar coordination for rX /rA ≥ 0.15, tetrahedral coordination for rX /rA ≥ 0.22, sixfold octahedral coordination for rX /rA ≥ 0.41, and eightfold coordination for rX /rA ≥ 0.73. Adapted from Hammond (2006).

These stacking sequences not only describe the crystal structures of different elements, but they also describe the structure of a wide range of compounds with two or more elements. Specifically, they describe compounds where smaller atoms (cations) fit into the interstitial spaces between larger atoms (anions). The available interstitial sites, with fourfold coordination, called tetrahedral sites, and sixfold coordination, called octahedral sites, in FCC and BCC crystals are shown in Figure 4.1.12a–d. From simple space filling considerations, assuming that the atoms are hard spheres, the radius ratios, rX /rA , determine the size, X, of the second atom/ion that can be accommodated in these interstitial sites. In the FCC structure, rX /rA = 0.225 and rX /rA = 0.414, for the tetrahedral and octahedral sites, respectively. It is easy to calculate the radius ratios for these sites based on the hard sphere model. Consider the tetrahedral site in the FCC structure shown in Figure 4.1.12b. The the atoms are in contact gives 4r √ face diagonal of the cube along which √A = cube ( 3a) 2a; the interstitial atom is located at 1/4th the body diagonal of the √ 3a/4. Thus and this distance is the sum of the two radii, i.e. rX + rA = rX /rA = 0.225. In the BCC structure, the octahedral sites are located both at the center of the faces and at midpoints between the edges (Fig. 4.1.12c), whereas the tetrahedral sites (Fig. 4.1.12d) are located halfway in between on a line joining the centers of the faces and the midpoints of the edges. Moreover, the octahedron is somewhat squashed in one direction as shown. Thus, in the

OUP CORRECTED – FINAL

236

Crystallography and Diffraction BCC structure, rX /rA = 0.291 and rX /rA = 0.154, for the tetrahedral and octahedral sites, respectively. The sizes of the interstices are practically important; for example, in carbon steels, carbon occupies the tetrahedral site in BCC iron, but since it is much larger (rC /rFe > 0.29), it distorts the structure and strengthens the steel.

4.1.6 Describing Crystal Structures—Some Examples In general, we are interested in describing the structures of materials or compounds of different elements; in particular, we wish to specify the coordinates of the basis atoms in the Bravais lattice as a way to calculate the intensities of diffraction patterns (discussed further in §7.4). In order to do this correctly, the arrangements of atoms in the unit cell must follow two simple rules: 1) First, if the crystal structure has a body, face, or base centering translation, it must begin and end on the same atomic species (see later discussion of CsCl). For example, such translations are (000, ½ ½ ½), referred to as body centering, in BCC crystals, and (000, ½ ½ 0, ½ 0 ½, 0 ½ ½), referred to as face centering in FCC crystals. In simple terms, the Bravais lattice must be populated in such a way that these centering translations move one atom, A, to another atom, A, of the same kind in the unit cell. 2) Second, the Bravais lattices and the actual crystals generated from them, possess various kinds of symmetry. By symmetry, we mean that there exist certain operations, such as reflection, rotation, inversion, translation, etc., called symmetry operators, that when applied to the crystal renders its structure unchanged. Further, each crystal system possesses a characteristic symmetry element (Table 4.1.1, column 4); we discuss symmetry in more detail in the next section (§4.1.7). However, it is important that each atom type in the crystal must independently satisfy the symmetry requirements of the entire crystal (see later discussion of NaCl). Simply put, any symmetry operation particular to the crystal system must move an atom, A, into coincidence with another atom, A. In other words, the symmetry operation cannot move an atom, A, to a different atom, B. We apply these two rules by first considering the crystal structures of Cu (FCC) and V (BCC). These two structures are cubic, with atomic positions in the unit cell, given by 4 Cu at 000, ½ ½ 0, ½ 0 ½, and 0 ½ ½, or, 4 Cu at 000 + face centering translations and 2 V at 000 and ½ ½ ½, or 2 V at 000 + body centering translations. Next, we consider the structures of the two alkyl halides, CsCl and NaCl. The unit cell of CsCl is shown in Figure 4.1.13a, with Cl– ions at 000, and Cs+ ions at ½ ½ ½. The Bravais lattice is simple cubic, and not BCC, because in the latter,

OUP CORRECTED – FINAL

The Crystalline State 237 (a)

(c)

a3

(e)

a3

a3

a2

a2

a1

a1 Cl–

(b)

Cs+

a2 a1 Ca2+

C (or Si)

(d)

a3

F–

(f)

2

1

3

[010]

A B

a2

a1 Cl–

Na+

S

Zn

S

Zn

Figure 4.1.13 Unit cells of typical crystal structures. (a) CsCl with two interpenetrating simple cubic unit cells (b) NaCl with two interpenetrating FCC unit cells. (c) The diamond cubic structure of carbon and common to elemental semiconductors (Si, Ge,). (d) The zinc-blende structure common to many compound semiconductors, including GaAs, ZnS, etc. (e) The CaF2 structure, which is a variant of the diamond cubic structure. (f ) The hexagonal Wurtzite structure.

the body centering translation would violate our first rule, and take a Cl– ion to a Cs+ ion. In the language of crystallography, we have a simple cubic lattice with a basis of two ions, Cl– ion at 000 and a Cs+ ion at ½ ½ ½. Similarly, one can interpret the NaCl, or the rock salt structure, shown in Figure 4.1.13b. Here it is easy, following (4.1.1), to see that the unit cell contains 4 Cl– ions at 000, ½ ½ 0, ½ 0 ½, and 0 ½ ½, and 4 Na+ ions at ½ ½ ½, 00 ½, 0 ½ 0, and ½ 00. The Cl– ions are face centered cubic, as are the Na+ ions, but with interpenetrating lattices displaced by ½ ½ ½. Thus, the overall Bravais lattice of NaCl is FCC, and the positions of the ions in the unit cell can also be written as 4 Cl– ions at 000 + face centering translations, and 4 Na+ ions at ½ ½ ½ + face centering translations. Further, in agreements with the second rule, the application of the characteristic symmetry operation for the cubic system, i.e. the threefold rotation

OUP CORRECTED – FINAL

238

Crystallography and Diffraction axes, or 120◦ rotation, along , as shown, leaves the unit cell unchanged (for example, look at Cl– ions, marked as 1, 2, 3, in the figure). Note that many monoxides, MO, where M2+ , a metal ion, replaces the Na+ , and O2– replaces the Cl– in the rock salt structure, such as MgO, FeO, NiO, etc., are well known. Semiconductor materials come in crystal structures that are variants of the FCC structure; in fact, the structures of all the well-known covalently bonded semiconductors can be derived from a parent FCC lattice with the cations filling the available tetrahedral (and octahedral) sites [1] (see also Exercises: Problem 13). Two structures commonly observed in semiconductors—diamond cubic (e.g. Si, Ge) (Fig. 4.1.13c) and zinc-blende (e.g. GaAs, SiC, ZnS, HgI) (Fig. 4.1.13d)—-are both derived from the FCC structure. The structure of diamond, with eight atoms per unit cell (Fig. 4.1.13c) is similar to Cu, discussed earlier, but with two atoms forming the basis, with coordinates (for carbon) of 4 C at 000 + face centering translations, and 4 C at ¼ ¼ ¼ + face centering translations. Notice that ¼ ¼ ¼ is a tetrahedral site in the FCC unit cell described in Figure 4.1.12b. The zinc-blende (ZnS) structure, also known as sphalerite, is similar to diamond cubic, but with a different atom on the two sites: 4 S at 000 + face centering translations and 4 Zn at ¼ ¼ ¼ + face centering translations. ZnS is also found in the wurtzite structure (Fig. 4.1.13f), with a hexagonal primitive lattice and atom positions: 2 Zn at 00 ½ + z, and 2/3 1/3 z, and 2 S at 000, and 2/3 1/3 1/2. Here, z, is a variable parameter with a value of z ∼1/8. Another variant of the diamond cubic structure is the CaF2 structure (Fig. 4.1.13e). Here the Ca2+ ions form an FCC unit cell, where the F– ions occupy ALL the eight available tetrahedral sites, with the positions 4 Ca2+ at 000 + face centering translations; 4 F– at ¼ ¼ ¼ + face centering translations, and 4 F– at ¼ ¼ ¾ + face centering translations. One complex-oxide structure observed in technologically important materials, built on the cubic lattice, is perovskite (ABO3 ), where A and B are cations occupying the body center (NI = 1) and corners (NC = 1), respectively, and the oxygen occupies the face centers (N F = 3), consistent with its stoichiometry. Other complex-oxide unit cells of interest are spinel structure compounds, (MO. Fe2 O3 ), introduced at the beginning of this chapter (frontispiece), where M is a divalent metal, with eight formula units or 56 atoms per unit cell, and 3+ 3+ garnets M3+ 3 Fe2 Fe3 O12 , with 160 atoms per unit cell! The spinel structure is also based on an FCC Bravais lattice, and hence the number of atoms per unit cell for each element, as expected, is a multiple of four. Moreover, in spinels, two formula units of 14 atoms are associated with each point of the FCC Bravais lattice.

OUP CORRECTED – FINAL

The Crystalline State 239 Example 4.1.4: An oxide of Cu has the structure, a0 = b0 = c0 = 4.27 Å, α = β = γ = 90◦ , with basis atoms Cu: ¼ ¼ ¼, ¾ ¾ ¼, ¾ ¼ ¾, ¼ ¾ ¾ and O: 000 and ½ ½ ½. (a) Draw a projection of the unit cell on the a,b plane, as well as the unit cell. (b) What is the stoichiometry (chemical formula) of this compound? How many formula units are in a unit cell? (c) What is the shortest Cu–O distance? (d) What is the density of this compound? Solution: (a)

O Cu

(b) Cu2 O; there are two formula units in the unit cell. √

(c) From the figure, the shortest distance is (d) The density of the compound is ρ = 6,153

3a0 ˚ 4 = 1.85 A. 2(2 × 63.54 + 16)1.674 × 10−27  3 4.27 × 10−10

=

kg/m3 .

4.1.7 Symmetry and the International Tables for Crystallography So far, we have described crystal structures in terms of a lattice and basis of atoms, within the constraints of simple rules of space filling. There is an alternative, more rigorous, and systematic way to describe how repetitive patterns can fill space. By assigning atoms, or molecules, or groups of atoms and molecules to these patterns, subject to symmetry constraints, we can arrive at a general description of crystal structures. We present a brief description of this approach, beginning with the essential ideas in two dimensions; the extension to three dimensions will follow logically, but only the basic principles are introduced in this chapter. Our goal here is twofold; to introduce the characteristic (minimum) symmetries of

OUP CORRECTED – FINAL

240

Crystallography and Diffraction

5 A revised version of the original publication, The International Tables of X-ray Crystallography, by N. F. M. Henry and K. Londsdale (1952).

the seven different crystal systems (Table 4.1.1, column 4) and the International Tables for Crystallography,5 of use in describing all the periodic crystal structures and interpreting diffraction patterns (§7 and §8). We define an asymmetric object as one with no remaining symmetry, such as the shape of a hand, or the letter, R. By subjecting the asymmetric object to two symmetry elements or operations, i.e. a mirror, m, and different rotation axes, we can evaluate the patterns that are generated. In conventional crystallography, a fivefold rotation axis is not encountered, as it is not consistent with lattice translations (see Fig. 4.4.1). Neither, do we encounter rotation axes greater than sixfold. Thus, the rotation axes of interest are onefold or rotation by 360◦ (represented by the symbol, 1), twofold (diad) or rotation by 180◦ (represented as 2), threefold (triad) or rotation by 120◦ (3), fourfold (tetrad) or rotation by 90◦ (4), and sixfold (hexad) or rotation by 60◦ (6). Each one is also represented by the number, n, where 360◦ /n represents the rotation angle. The combination of a mirror with each of these rotation axes gives rise to 10 crystallographic point groups in two dimensions (also known as plane point groups) (Fig. 4.1.14). They are called point groups because their rotation axis, perpendicular to the page, and the mirror line(s) pass through a single point. Furthermore, the point groups correspond to the two-dimensional motifs, or projections of molecules in two dimensions (Fig. 4.1.14). What kinds of lattices are consistent with these point groups, or how can we extend the symmetry elements of the motifs in a two-dimensional pattern? Rotation axes with n = 1, 2, can be repeated in a pattern consisting of a general parallelogram ( p1 and p2 in Fig. 4.1.15). Since the mirror symmetry must extend through the whole pattern, when a mirror is introduced, the parallelogram is reduced either to a simple ( pm) or centered (cm) rectangular lattice. To be consistent with a fourfold rotation axis (n = 4), the lattice has to be a square ( p4), and n = 3, 6 generate an equilateral triangular ( p3 and p6) lattice. In all, for two-dimensional crystals, we have the possibility of five lattices (Fig. 4.1.15). In addition, the periodic patterns, combining the allowed rotation axes (1, 2, 3, 4, and 6) and the translations, must be generated from only these five possibilities. It turns out that there are only 10 periodic patterns in total. Now, the addition of point groups containing mirrors with the lattice translations gives rise to a new symmetry element—the glide mirror, which is combination of a mirror and a translation. When the glide is combined with the five lattices, it generates seven more periodic patterns, giving a total of seventeen two-dimensional patterns, or plane groups (also known as two-dimensional space groups; Fig. 4.1.15). To summarize then, in two dimensions, we have 10 crystallographic point groups or motifs (Fig. 4.1.14), five possible lattices, and 17 different plane groups (twodimensional space groups) or patterns. To extend this analysis to three dimensions, we have to introduce three additional symmetry elements: inversion (which transforms all points xyz to xyz), roto-inversion (combination of rotation and inversion), and screw (combination

OUP CORRECTED – FINAL

The Crystalline State 241

1

CI C Br

R

C

R R

H

F C H

F

bromochlorofluoroethene

H

R

H H C C H H ethene

m

R m

R

H

F C

H N

C

H F

R

R

F WOF 4

m m m

R

R

m

RR

R R R

R R (6) - rotane, C18H24

W

F

m

4mm

F tungsten oxyfluoride H C

m

RR

R

R R

F

H

m

R RR R

R

3m

H boric acid

RR

R

6

2mm

H

O

m

(4) - rotane, C12H16

R

B O

RR RR

R

O

m

R

H

R

H H trifluoralkylammonia

m

RR

R

C

m

RR

H

F

4

m

trans-difluoroethene

R

3

C

F R

2

H C F

H

cis-difluoroethene

m

R R

R

F

C

C

H C

m

C H C

m

H

6mm

C

benzene

Figure 4.1.14 The 10 crystallographic point groups in two dimensions or plane point groups, including examples of molecules with the same symmetry. On the left are the pure rotations and on the right are possible combinations of rotations and mirrors. Note that a mirror (m), combined with a rotation axis generates an additional mirror for the twofold (2mm), fourfold (4mm), and sixfold (6mm), but does not for threefold (3m) rotations.

H

of rotation and parallel translation). In addition, combinations of two rotation axes generate a third rotation axis passing through the same point. However, to be compatible with lattice translation, the generated axis can only be one of the five possible rotation axes (1, 2, 3, 4, and 6). The latter restricts the rotation axes combinations to the monoaxial group (simple rotations of 1, 2, 3, 4, and 6), dihedral group or sets of two twofold rotation axes combined with a third (222, 223, 224, 226), the isometric groups, 233 (tetrahedral) or 234 (octahedral), and the special case, 4 (an inversion tetrad axis). Furthermore, in three dimensions, combinations of mirrors and translations give rise to a number of possible glide planes, and combinations of rotations and translations give screw axes, not encountered in two dimensions. Including these symmetry elements in three dimensions, we find a total of 32 crystallographic point groups (Table 4.1.1), some without (column 5) and some with (column 6) a center of symmetry. This is an important distinction as a diffraction pattern is always centro-symmetric, even if the crystal does not possess a center of symmetry; this phenomenon is known as the Friedel law (to be discussed further in §7.6.2). Furthermore, the five possible plane lattices in two dimensions expand to the 14 Bravais lattices presented in Figure 4.1.4. Moreover, the 32 point-groups when mapped into the seven crystal systems,

Adapted from Hammond (2006).

OUP CORRECTED – FINAL

242

Crystallography and Diffraction

p1

Figure 4.1.15 The 17 twodimensional space groups (also known as the plane groups) of importance in surface electron diffraction (see §8.3). Note that these are based on the five possible plane lattices: the oblique p-lattice (p1, p2) with a  = b, γ = 90◦ ; the rectangular p-lattice (pm, pg, p2mm, p2mg, p2gg) with a  = b, γ = 90◦ ; the rectangular c-lattice, allowing for an additional centering translation and creating a multiple cell (cm, c2mm); the square p-lattice (p4, p4mm, p4gm) with a = b, γ = 90◦ ; and the hexagonal p-lattice (p3, p6, p3m1, p31m, p6mm) with a = b, γ = 120◦ . Adapted from Hammond (2006).

p2

pg

cm

p2mm

p2mg

p2gg

c2mm

p31m

p3m1

p3

p6mm

p6

p4

pm

p4mm

p4gm

reveal important characteristic symmetries associated with each crystal system (Table 4.1.1, column 4). For example, the characteristic symmetry for the cubic system is a set of four triad (n = 3) axes, equally inclined at 109.47◦ along the body diagonal. Similarly, the orthorhombic system, is characterized by three diads (n = 2), equally inclined at 90◦ or equivalently, a set of three orthogonal mirrors. Finally, the 17 plane groups or patterns in two dimensions, increase to 230 space groups in three dimensions, arising from the combination of the fourteen Bravais lattices with the 32 point-groups. These 230 space groups are numbered, beginning with triclinic (lowest symmetry) and ending with cubic (highest symmetry), drawn systematically, and described in the International Tables for Crystallography (ITC). Note that to be proficient in materials characterization, it only requires an ability to read and interpret these tables.

OUP CORRECTED – FINAL

The Crystalline State 243 Space group Symbol

Replacing the translation (n,a) symbols by mirrors gives crystal class mmm - orthorhombic system Full symbol, includes redundant 21 screw axis

created by mirror and glide planes

y x

Schoenflies Notation a-glide⊥ to z-axis at c/4 above the plane Space group in terms of symmetry elements n-glide⊥ to x-axis

Space group in terms of asymmetric unit Represents the effects of symmetry elements

General positions Special positions

Mirror⊥ to y-axis

Conditions limiting possible reflections or systematic absences in diffraction

Figure 4.1.16 An annotated page from the International Tables for Crystallography, Space group #62, Pnma. Adapted from Allen and Thomas (1999).

Figure 4.1.16 shows an annotated example of a page from the ITC, for space group Pnma (#62). Each space group includes two drawings of the unit cell, typically projected in the x-y plane, one representing all the symmetry elements, and the other, the equipoints in the cell generated by applying the symmetry elements (shown in the second drawing) to an asymmetric object (◦)—defined as any object with no remaining symmetry—placed in the asymmetric unit. Note, that a mirror image is indicated by . Further, this is an orthorhombic crystal system as it involves the characteristic symmetry of three orthogonal mirrors, mmm. The coordinates of the general positions, as well as their number (8) per unit cell, the “Wycoff letter” identifying the position (d), and the symmetry (1) of the position are also shown under the drawings. If the asymmetric object were to be placed in a special position, such as the mirror, a simpler pattern would result. The eight asymmetric units would merge into four, and occupy the special equivalent position, and described as 4c. Similarly, the special positions 4a and 4b, both with multiplicity of four in the unit cell, represent sites with inversion symmetry (1). Note that any atom or molecule occupying a special position must at least possess the minimum symmetry of the site. Last, but most importantly, from the point of

OUP CORRECTED – FINAL

244

Crystallography and Diffraction diffraction, the table on the bottom right lists the Laue conditions, or when one is expected to observe systematic absences of certain reflections in X-ray diffraction intensities. We discuss this more in §7, where we cover X-ray diffraction intensities in a quantitative manner. We now revisit some of the structures discussed in the last two sections and assign their space groups and numbers from the ITC. Simple cubic metals are Pm3m (#221), FCC metals are Fm3m (#225), BCC metals are Im3m (#229), and HCP crystals are P63 /mmc (#194). CsCl is also Pm3m, but with both Cs and Cl occupying special positions. The structure has a center of symmetry, as indicated by 3, the triad roto-inversion axis. The zinc blende structure of the semiconductors ZnS and GaAs is the space group, F43m (#216), with atoms in special positions; note the absence of the center of symmetry or no inversion. The diamond cubic structure is Fd3m (#227). Only a very small number of space groups are commonly observed, and many of the 230 possible space groups do not correspond to any actual crystal structures. Most of the elements (70%) belong to five space groups (Fm3m, Im3m, Fd3m, F43m, and P63 /mmc), and 60% of the organic/inorganic compounds belong to six space groups (P21 /c, C2/c, P21 , P1, Pbca, P21 21 21 )! Not surprisingly, these space groups are the ones that lead to close packed crystals.

4.1.8 The Stereographic Projection Crystals are three-dimensional objects and in any written text, such as this one, we need to represent them on a two-dimensional page. While this can be done in combinations of plan and elevation drawings, the angular relationships between lattice planes and directions are better represented in the stereographic projection. Figure 4.1.17 shows the principles of this projection. The crystal of interest is placed at the center of the sphere, normals to all planes of interest are drawn, and their intersections with the surface of the sphere, called poles, are recorded (Fig. 4.1.17a). Note that the angle between the poles, φ, is different from the dihedral angle, , between the faces (Fig. 4.1.17c); however, φ +  = 180◦ , is always true. Moreover, for any two poles on the surface of the sphere, there always exists a great circle that passes through both of them. To complete the stereographic projection, lines are drawn from the poles to the south pole (Fig. 4.1.17b), and the intersection of these lines with the equatorial plane constitutes the stereographic projection (Fig. 4.1.17b). The stereographic projection (Fig. 4.1.18a) has some specific properties: (a) Any circle on the sphere also appears as a circle on the plane of projection. However, the centers of the two circles, generally, do not coincide. (b) A great circle on the sphere appears as a circular arc on the projection. Further, it cuts the basic circle of the projection at two, diametrically opposite, endpoints.

OUP CORRECTED – FINAL

The Crystalline State 245 (a)

N

(b)

N

(c)

F1



No r of mal fac e

Great circle

F2

φ

S

S



Figure 4.1.17 Basics of the stereographic projection. (a) The crystal (in this case, Galena, the cubic form of PbS) is placed at the center of a unit sphere. The normal to the crystallographic planes are extended and their points of intersection, called poles of the planes, are recorded. (b) Lines are drawn from each pole to the south pole, S, of the sphere as shown. The intersection of this second set of lines with the equatorial plane (grey) is the stereographic projection. A detailed example, for a complete cubic crystal, is shown in Figure 4.1.18. (c) Schematic illustrating the angle, φ, between the normals, as well as the dihedral angle, , of two planes, F1 and F2 . (c) Angles are preserved in this projection. Thus, it is possible, to measure the angle of intersection between great circles on the projection. This is done with a Wulff net (see Fig. 4.1.18b and the discussion that follows). (d) Unfortunately, areas are not preserved in the projection. In practice, the stereographic projection is used in conjunction with a Wulff net (Fig. 4.1.18b). The planes and directions are first plotted in a stereographic projection and then this is overlaid on a Wulff net, such that the center of the stereographic projection and the Wulff net are superimposed and they can rotate relative to one another. Now, with this arrangement it is possible to do the following: (a) Measure the angle between two crystallographic planes or directions. The Wullf net is rotated such that the stereographic projection of the two direction of interest lie on the same great circle. Then the angles are simply read off the Wulff net. (b) To rotate the projection about an axis, align the axis of rotation with the N–S axis of the Wulff net. Then each point of the projection is moved by the required angle of rotation along the small circle of the Wulff net on which it lies. (c) To index unknown crystallographic directions, compare the angle between them and known crystallographic directions, with the angle between indexed directions in a standard projection.

OUP CORRECTED – FINAL

246

Crystallography and Diffraction (a)

– 100

–– 510

–– 310

– 510

(b)

– 310

–– 210

––

110

–– 321

–– 311

–– 312

–– 231

–– 212

––

– 313

– 221

– 212

–– 130

–– 121 –– 131

– 021

–

010

–

011 – 132

– 131

– 150

– 122 – 121

– 130

–– – – 113 123 – 013 – 012 – – – 123 113 133

– 212

– 221

–

– 211 – 321

110 – 210

– 112

011

021

010

132

133

103

131 112

150

122

213

130

121

111

313 212

– 312

– 150

013

113 123

– 313

– 130 – 131

– 132

013

231

101

332

201

120

221

311

312

– 311

– 310

– 122

– 133 – 123

– 113

001 102 – 213

– 113

– 111

– 231

– 120

– 102 – 103

–– 133

–– 132

– 121

– 213

–– 213

–– 112

–– 122

– 120

– 231

– 111

111 –– 150

110

– 321

– 312

–

101

–– 313

–

– 311 – 211

– 201

–– 211

–– 221

–– 120

N

– 210

321

311

110

210 – 510

100

510

310

S Figure 4.1.18 (a) A complete stereographic projection of a cubic crystal. (b) A Wulff net used for measuring angles (see text for details). Finally, for those interested in reviewing the use of stereographic projections and Wullf nets in diffraction, it is worth their while looking up the detailed treatment of the subject in Cullity (1978). Example 4.1.5: Construct the stereographic projection for the point group 4mm shown in Figure 4.1.14. Solution: We construct it in parts as follows: (a)

(b)

(c)

(a) Fourfold rotation axes showing the equipoints. (b) Mirror planes, shown as bold lines, perpendicular to [100]. The fourfold rotation axes also act on the mirrors, such that two additional mirrors perpendicular to [110] are also generated.

OUP CORRECTED – FINAL

The Crystalline State 247 (c) However, the new mirrors perpendicular to [110] do not generate any additional equipoints in the stereographic projection.

4.1.9 Imperfections in Crystals So far, we have discussed ideal crystalline materials; in practice, ideal crystals are seldom found, and real crystals show distinct features, called defects, that deviate from the ideal. Such defects determine materials behavior as many of their properties are profoundly affected by deviations from the ideal crystalline structure. Hence, evaluating such imperfections is central to materials characterization and analysis. Imperfections are classified based on lattice irregularities as point, line, and planar defects. There is a vast literature on the subject, including some classic textbooks (Nabarro, 1967; Weertman and Weertman, 1992), but we only present an introduction that is relevant for a description of their analysis in later chapters (§9.5.2). 4.1.9.1 Point Defects The simplest or point defect is a vacancy in a lattice site originally occupied by an atom. Vacancies are found in all crystals (Fig. 4.1.19). Alternatively, a selfinterstitial is an atom from the regular crystal that is found elsewhere at an interstitial site; the latter, typically of a smaller volume than a regular site. Often, in metals, such interstitials lead to large distortions and strain in the surrounding lattice. If a second element, either an impurity or a deliberate addition, as in alloying, is included, it can either be a substitutional atom, forming a solid solution, or occupy an interstitial position. The incorporation of an additional element as a solid solution will be determined by (a) atomic size ratios to minimize distortion of the lattice, (b) crystal structure—-preferably the same for the element and alloy, (c) electronegativity difference—to be as small as possible, and (d) their valence. Accurate characterization of point defects is difficult and challenging, but most success has been obtained by positron annihilation spectroscopy (PAS), a technique that is beyond the scope of this book. However, for those interested, an excellent introduction to the subject of PAS, emphasizing its applications to metals and alloys, can be found in this accessible review [2]. 4.1.9.2 Line Defects This type of linear defect, called a dislocation, is often associated with local atoms that are misaligned in some way. An edge dislocation (Fig. 4.1.20a) involves an extra half-plane of atoms, referred to as the dislocation line, which terminates within the crystal. The extra half-plane of atoms leads to a displacement of a part of the crystal with respect to the rest. For an edge dislocation, this displacement vector,

OUP CORRECTED – FINAL

248

Crystallography and Diffraction

Interstitial

Figure 4.1.19 Point defects are of two kinds, interstitial or vacancy, as shown.

(a)

Burgers Vector b

Vacancy

(b)

(c)

Edge Dislocation Line Dislocation Line Burgers Vector Figure 4.1.20 Schematic illustration of (a) edge and (b) screw dislocations; note that, in the latter, all the atoms are displaced parallel to the Burgers vector, b. See §8.5.4 and §9.5.2. In reality, dislocations tend to have partial edge and partial screw character. (c) One of the earliest TEM images of dislocations in cold-rolled stainless steel. Adapted from [5].

known as the Burgers vector, b, is normal to the dislocation line. A second type of dislocation, called the screw dislocation, arises from shear stress to produce a linear displacement of atoms (Fig. 4.1.20b). Screw dislocations are characterized by the Burgers vector, b, being parallel to the dislocation line, and play an important role in crystal growth. Edge and screw dislocations are end-members of a continuum, as most dislocations found in crystalline materials are mixed, having both edge and screw components. From a practical point of view, deformation of crystalline materials involves the motion or pinning of dislocations, and as such, they play a very important role in plastic deformation in materials. Dislocations are observed and their Burgers vector determined by electron microscopy techniques (Fig. 4.1.20c) and further details are in §9.5.2.

OUP CORRECTED – FINAL

The Crystalline State 249 4.1.9.3 Planar Defects Figure 4.1.1 introduced the concept of grain boundaries. Figure 4.1.21a schematically shows the atomic mismatch between two crystalline grains, and Figure 4.1.21b shows an atomic resolution image of the same. A small-angle grain boundary involves a small mismatch and can be described by an array of dislocations. If the grain boundary is formed by a series of edge dislocation (Fig. 4.1.21c), it is known as a tilt boundary. If the grain boundary is formed by a series of screw dislocations, it is known as a twist boundary. The interatomic distances at a grain boundary are typically larger than that of an ideal crystal of the same material; as such, they are higher-energy arrangements of the atoms and the magnitude of their interfacial energy is proportional to the degree of misalignment. A stacking fault is typically found in an FCC crystal, where the ABCABC . . . stacking of the close-packed planes (§4.1.5) is interrupted and locally altered to an ABCABABABC . . . stacking. Similar changes can also be expected to occur in HCP crystals, i.e. local changes from ABABAB . . . to ABABCABA . . . stacking. A twin boundary is a special kind of grain boundary arising from the growing together of two crystals with well-defined orientation relationships. A common twin element is a mirror (Fig. 4.1.21d), such that atoms on one side of the boundary are located in positions that are mirror images of the atoms on the other side. Twins are formed during growth or through mechanical deformation, and occur on well-defined crystallographic planes; the characteristic twin planes are {111} and {112} for FCC and BCC, respectively (see §8.7.6 for further discussion of twinning). An antiphase boundary (APB), is the interface between two domains in an ordered alloy. Such interfaces are typically coherent and across the boundary the lattice planes are continuous but the site occupancy is different. A good example is the APB in B2 ordered alloys with a BCC unit cell, where on one side atoms (a)

(d)

(c)

Figure 4.1.21 (a) Schematic representation of a grain boundary (b) An atomic resolution microscopy image of a GB in Mo. The white dots are separated by 0.23 nm. (c) Schematic of a small angle tilt boundary consisting of a series of edge dislocations. (d) A grain boundary in Al. The white dots are separated by 0.2 nm.

(b)

θ

Adapted from Williams, Pelton and Gronsky (1991).

OUP CORRECTED – FINAL

250

Crystallography and Diffraction of type A occupy the cube corners and atoms of type B occupy the body centers, and on the other side of the APB the positions of atoms A and B are reversed. Defects are analyzed by various imaging techniques that include transmission electron diffraction (§8.6) and microscopy (§9.5.2). They also do alter positions of atoms in the crystal, and this in turn, affects the intensities and positions of observed reflections in diffraction. We discuss this further in §8. The problem of crystal structure determination by diffraction can be largely divided into two parts. The first part is to obtain the size and shape of the unit cell, including the lattice parameters, from the geometry of the diffraction pattern. The second, much more challenging, part is to obtain the lattice type, and the positions of the atoms in the unit cell. Section 4.3 introduces the concept of diffraction as a way to provide an overview of the first part, and leave the details, as well as the second part, to later chapters, i.e. §7 and §8. But first, to understand diffraction, we now describe the important concept of the reciprocal lattice.

4.2 The Reciprocal Lattice

6 The factor 2π is used for the proportionality constant, K, to be consistent with our definition of the magnitude of the wave vector, | k | = k = 2π/λ (see §6 on optics and optical methods, and specifically, §6.2 where k is defined). Alternatively, we can set K = 1, but then we have to redefine the wave vector | k | = k = 1/λ, which is the practice in some textbooks. Note that K = 1 is commonly used in the interpretation of high-resolution electron microscopy. See §9.2.7.3 and specifically (9.2.17).

X-ray diffraction was originally interpreted as an interference effect of X-rays reflected by a parallel set of lattice planes that depends on both their orientation and their interplanar spacing[2]. From this perspective, the reciprocal lattice construction is an excellent aid, as each of its lattice points represents an entire family of lattice planes, (hkl), at a given orientation and a fixed interplanar spacing, dhkl . Needless to say, the reciprocal lattice has a fixed relationship to the real lattice, from which it is derived. As such, the reciprocal lattice can be constructed using either a physical or a mathematical (vector) approach, and we now begin with the former. Consider the three families of planes, with inter-planar spacing d 1 , d 2 , and d 3 , perpendicular to the plane of the paper (Fig. 4.2.1a). We draw normals to the planes (Fig. 4.2.1b) and set the vectors (Fig. 4.2.1c) proportional to the inverse of their inter-planar spacing, i.e. | d∗i | = K/di , where the proportionality constant, K = 2π, is assumed.6 The vectors, d∗i , i = 1, 2, 3, are called the reciprocal lattice vectors, as their dimensions are 1/length (in general, Å–1 ). Notice that d∗3 = d∗1 + d∗2 , and these three vectors, can indeed generate a space lattice. A similar exercise can be done with the x-z projection of a primitive monoclinic unit cell, with the b axis normal to the page (Fig. 4.2.2a). Figure 4.2.2b shows the corresponding reciprocal lattice vectors for many of the planes, and Figure 4.2.2c shows the projected unit cell of the reciprocal lattice. It is clear that the magnitudes ∗ ∗ of the reciprocal lattice vectors are | a∗ | = d100 = 2π/d100 and |b∗ | = d010 = 2π/d010 . We can extend this approach to the third dimension and generate the complete unit cell of the monoclinic P crystal in reciprocal space (Fig. 4.2.5a).

OUP CORRECTED – FINAL

The Reciprocal Lattice 251 (a)

(b)

planes 3

(c)

planes 2

d2 d3

normal to normal to planes 3 planes 2

planes 1

d1

d1*

normal to planes 1

d*2

d3*

Figure 4.2.1 The reciprocal lattice construction. (a) A schematic of three sets of lattice planes in real space for a monoclinic crystal. (b) Directions of the normal to each of the sets of planes. (c) The reciprocal lattice vectors, di , with magnitudes inversely related to their inter-planar spacing. Note that d∗3 = d∗1 + d∗2 .

(001) (002)

a

) 01

c

(c) 102

)

02

(1

β O

d*102 d*101 d*100

(002)

d*101

002

d*002 101 d*001 O d*001

001 c*

100 a* 101

000 β* 001

Figure 4.2.2 (a) A monoclinic unit cell, with the b axis normal to the page (x-z projection), and some planes are indexed as shown. (b) The reciprocal lattice vector for some of the planes shown in (a). Note that the reciprocal lattice vectors are oriented normal to the planes and their lengths are related inversely to their inter-planar spacing. (c) The unit cell of the reciprocal lattice. Adapted from Hammond (2006).

1)

0 (1

(002)

(b)

(1

(a)

(100)

Adapted from Hammond (2006).

Now that we have a physical idea of the reciprocal lattice and how it is generated, c* we can take a look at the mathematical approach that is easier to generalize. We start with the most general case, or the lowest symmetry triclinic unit cell P c b B defined by the vectors, a, b, and c (Fig. 4.2.3). Then the volume, V, of the unit C b* cell is the area of the parallelogram, OACB, multiplied by the height of the cell, a A OP. In vector notation, we can write this as V = a · (b ⊗ c) = b · (c ⊗ a) = O a* c · (a ⊗ b). Then the reciprocal lattice vectors, a*, b*, and c* (Fig. 4.2.3) are defined as Figure 4.2.3 The unit cell of a primitive triclinic unit cell showing the lat∗ a = 2π (b ⊗ c)/V tice vectors in real space as well as the (4.2.1) b∗ = 2π (c ⊗ a)/V reciprocal lattice vectors. c∗ = 2π (a ⊗ b)/V

OUP CORRECTED – FINAL

252

Crystallography and Diffraction 1. 57 Å–1 (a)

(b) 1Å

020

120

220

(010) b 010 Figure 4.2.4 A simple cubic crystals showing (a) the real lattice with c-axis normal to the page, and (b) its reciprocal lattice. The unit cell parameter for the real crystal, a = 4 Å.

(110)

(100) b*

110

g110

210

g210

(210) c

a

c*

000

Crystal lattice

Adapted from Cullity (1978).

100 a* Reciprocal lattice

200

It is easy to see in the figure that c* is normal to the plane of a and b, and its length is 2π · area of parallelogram, OACB 2π | a ⊗ b | = V (area of parallelogram, OACB) x (height of cell, OP) 2π 2π = = , (4.2.2) OP d001

| c∗ | =

which is simply proportional (by a factor of 2π) to the inverse of the inter-planar spacing of the (001) planes of the real crystal. Similarly, we can show that a* is normal to the (100) planes, with magnitude, | a∗ |= 2π/d100 , and b* is normal to the (010) planes, with magnitude, | b∗ |= 2π/d010 . We can extend this approach to all the planes of the crystal lattice and build the reciprocal lattice by repeated translations of the reciprocal lattice vectors, a*, b*, and c*. We then label the array of points generated by this lattice in terms of the basis vectors, i.e. the end of a* is labeled as 100, the end of b* is labeled as 010, and the end of c* is labeled as 001. Hence, any vector, ghkl , in the reciprocal lattice, is normal to the set of planes with the Miller indices (hkl). Thus, ghkl can be written in terms of the unit cell vectors of the reciprocal lattice as ghkl = ha∗ + kb∗ + lc∗

(4.2.3a)

| ghkl |= 2π/dhkl

(4.2.3b)

In addition, its length7 is

7

If we set the proportionality constant, K = 1, we have |ghkl | = ghkl = 1/dhkl .

As a first exercise, Figure 4.2.4 illustrates the reciprocal lattice generated by the application of (4.2.1) to the unit cell of a simple cubic crystal. Note that g200 is, as expected, parallel to g100, but is twice as long. Also, as mentioned earlier, planes

OUP CORRECTED – FINAL

The Reciprocal Lattice 253 (a)

(b) 011

002

001 c* β*

112

c*

010

b*

000 a*

022

110

100

002

000 a*

222

222

202

111

101 b*

(c) 022

202

121

011 101 020 110

211 220

200

111 c*

b*

020 220

000 a* 200

Figure 4.2.5 Reciprocal lattice unit cells for the real (a) Monoclinic (b) BCC, and (c) FCC unit cells. Note that the reciprocal unit cell for the real BCC is FCC, and for the real space FCC the reciprocal unit cell is BCC. Further for the real BCC, the reciprocal unit cell satisfies the condition h + k + l = even, and for the real FCC structure, hkl must be either all even or all odd. The latter has implications for the intensities, including absences of specific reflections, observed in diffraction patterns (see §7.5). (nh nk nl), where n is an integer, are parallel to (hkl),but with 1/nth the inter-planar spacing. Thus |gnh nk nl | = n|ghkl |. The reciprocal unit cells of the cubic (I) unit cell (BCC), and the cubic (F) unit cell (FCC), are shown in Figure 4.2.5b and Figure 4.2.5c, respectively. The cubic unit cell in real space is based on mutually perpendicular basis vectors, a, b, and c. In such cases, which also includes the orthorhombic and tetragonal crystals, the reciprocal lattice vectors, a*, b*, and c*, are also mutually perpendicular. For crystals, such as monoclinic and triclinic, discussed earlier, where the unit cell vectors in real space are not mutually perpendicular, the construction of the reciprocal lattice involves complementary angles. As an example, Figure 4.2.6 shows the real and reciprocal unit cells of the hexagonal lattice. Note that in general, since c* is normal to both a and b, its scalar product with either one should be zero, i.e. c∗ ·a = c∗ ·b = 0 c∗ ·c = 2π

(4.2.4)

It is now easy to prove the validity of the zonal equation, (4.1.7), we discussed in §4.1.4. By definition, the planes of a zone are all parallel to a common line, [uvw], called the zone axis, and the normal, ghkl , to the plane must be coplanar and normal to [uvw]. Thus, the two vectors, the zone axis, Z = ua + vb + wc, and the reciprocal lattice vectors, ghkl = ha∗ + kb∗ + lc∗ , must be perpendicular with a scalar product of zero. Thus, Z · ghkl = (ua + vb + wc) · (ha∗ + kb∗ + lc∗ ) = 0 uh + kv + lw = 0 as illustrated in Figure 4.1.10.

(4.2.5)

OUP CORRECTED – FINAL

254

Crystallography and Diffraction 1. 57 Å–1

c* 000

(a)

(b)

1Å

b* b

c

a* 100

(120) Figure 4.2.6 The hexagonal crystal showing (a) the real lattice and the unit cell with c-axis normal to the page. The unit cell parameter ˚ for the real crystal, a = 4 A.(b) Its reciprocal lattice with some reciprocal lattice vectors indicated.

020

g110 110

(010)

(110)

010

g120

200

a

120

(100) 210 Crystal lattice Reciprocal lattice

Adapted from Cullity (1978).

220

Example 4.2.1: Consider an orthorhombic crystal with lattice parameters ˚ a = 2 A,

˚ b=3A

and

  ˚ α = β = γ = 90◦ . c=4A

(a) Draw the unit cell. (b) Calculate the inter-planar spacing, dhkl , using equation in Table 4.1.2, for the (100), (002), (030), (011), (111), (310), (311) and (312) planes. (c) What is the common direction for the (011) and (111) planes? Would either (310) or (311) planes, or both, share this common direction (i.e. form part of the same zone)? (d) Calculate the unit cell dimensions of the reciprocal lattice? Draw this in the same figure as (a) but with different scale (note: units are different). Solution: (a) The unit cell is shown in the following figure.

OUP CORRECTED – FINAL

The Reciprocal Lattice 255

z

c c* a

b

b*

y

a* x (b) The interplanar spacing can be calculated (Table 4.1.2), using 2 2 2 1 = ha2 + kb2 + cl 2 , as shown in the following table d2 hkl

Plane (100) (002) (030) (011) (111) (310) (311) (312) dhkl (Å) 2.0 2.0 1.0 2.4 1.54 0.65 0.64 0.62 (c) The common direction is given by the cross product, (4.1.8). Thus,   for (011) and (111), we have [uvw] = 011 . (311) would share the same common direction (dot product = 0), but (310) would not (dot product = 0). (d) The reciprocal lattice vectors are calculated using (4.2.1). Thus a∗ = 2π (b ⊗ c)/V = πi b∗ = 2π (c ⊗ a)/V = 2π/3 j c∗ = 2π (a ⊗ b)/V = π/2 k The reciprocal unit cell is also shown in the preceding figure (See Example 4.3.2).

We are now ready to introduce diffraction, both in real space and in reciprocal space formulations.

OUP CORRECTED – FINAL

256

Crystallography and Diffraction

4.3 Diffraction In 1912, von Laue8 made the earliest suggestion that X-rays might be “diffracted” by crystals. At that time, it was commonly understood that the geometry of macroscopic crystals could be explained by a three-dimensional arrangement of atoms with interatomic distances of the order of 1–2 Å. By comparing the relationship between visible light and the repeat distances of the rulings in gratings (Fig. 3.5.1) used in optical diffraction (gratings are discussed further in §6.6), with the wavelengths of X-rays (Fig. 1.3.2) and the periodicities in crystals, von Laue suggested that crystals could serve as three dimensional gratings for X-ray diffraction. Indeed, this prediction was immediately confirmed by the work of Friedrich and Knipping [3]. However, the simple and elegant picture of W. H. Bragg and W. L. Bragg quickly overshadowed this approach of considering the crystal as a three-dimensional grating. In the Bragg picture [4], the crystal is considered to be made up of planes of atoms that specularly9 reflect the X-rays. When the path difference between reflections from successive planes of the crystal of the “reflected” beam at a specific X-ray incidence angle (Fig. 4.3.1), is some integral number of the X-ray wavelength, they suggested that an enhancement in intensity, or “strong” beams are produced. Even though this picture of (a)

1

2 B´

Adapted from Cullity (1978).

A1

θ B

A2

C

2θ

(b) 1

d

2

d

A3 (c) 2

1

1´

θ

(d) 1 23

2´

θ

θ C

A B

1´3´ 2´

(100)

θ

AD F C E

(200)

Figure 4.3.1 (a) Schematic illustration of Bragg’s law of diffraction showing a simple scheme to calculate the path difference. (b) When the path difference for waves “reflected” from successive planes is an integral multiple of the wavelength, constructive interference, as shown, is expected. (c) A second-order diffraction from the (100) planes is equivalent to a first order diffraction (d) from the set of (200) planes.

θ

B

8 Max von Laue (1879–1960) was a German physicist who was awarded the 1914 Nobel prize in Physics for “his discovery of the diffraction of X-rays by crystals.” 9 Like a mirror.

OUP CORRECTED – FINAL

Diffraction 257 mirror-like reflection is, strictly speaking, physically incorrect, the geometry of scattering is correctly described by the simple Bragg’s law of diffraction.10

4.3.1 Bragg’s Law: Interpreting Diffraction in Real Space Consider a parallel, monochromatic beam of X-rays, incident at an angle, θ , on the surface of a crystal, with lattice planes separated by a distance, d (Fig. 4.3.1a). We assume that all three—the incident beam, the normal to the surface, and the diffracted beam, are always coplanar. Further, the diffraction angle, 2θ , between the incident and diffracted beams, typically measured in experiments, is always twice the incidence angle, θ. For diffraction or positive interference (Fig. 4.3.1b) the path difference between the two beams, 1 and 2, should be some integral number of wavelengths, nλ. The path difference, PD, between beams 1 and 2, in Figure 4.3.1a, is given by PD = BA1 − B A1 = BA3 − BC = CA3 = 2d sin θ. Thus nλ = 2d sin θ

(4.3.1)

where the integer, n, determines the order of the diffraction. We can rewrite (4.3.1) as  λ=2

dhkl n

sin θ

(4.3.2)

However, as we have seen before, dhkl = ndnh nk nl . Hence, n is not written explicitly, for we can consider a diffraction of nth order from any set of planes, as a first order diffraction from a parallel set of planes with 1/nth the spacing (Fig. 4.3.1c,d). Thus, Bragg’s law can be written in general terms as λ = 2d sin θ

(4.3.3)

and we will use (4.3.3) throughout the book. However, note that sin θ < 1, and ˚ the useful hence, λ < 2d, or 0 < 1/d < 2/λ. Since, in most crystals, d < 3 A, ˚ Further, if the incident range of X-ray wavelengths for diffraction is λ < 6 A. beam is along the x-axis, and the diffracting crystal is placed at the origin, O (Fig. 4.3.2), then all the allowed values of ghkl (= 2π/dhkl ) in reciprocal space, which have the potential to diffract X-rays with wavelength, λ, would be enclosed in a limiting sphere of radius, 4π/λ. In general, the interatomic spacing, dhkl , of the planes to be used in Bragg’s law, is a function of their Miller indices and the parameters (a, b, c, α, β, γ ) of the crystal lattice, as given in Table 4.1.2. As an exercise, we can start with the general equations for the triclinic crystal given in Table 4.1.2. Thus, for a tetragonal crystal, with axes a = b, and c, and α = β = γ = 90◦ , we get, V = abc = a2 c,

10 For those interested, Fifty Years of X-Ray Diffraction by Ewald et al. (1962) is an invaluable sourcebook covering the fascinating early beginnings of X-ray diffraction.

OUP CORRECTED – FINAL

258

Crystallography and Diffraction s11 = b2 c2 = a2 c2 = s22 , s33 = a2 b2 = a4 and s12 = s23 = s31 = 0. Substituting these values, we derive the expression in Table 4.1.2, for a tetragonal unit cell, and combine it with (4.3.3) to get 1 h2 + k2 l2 4 sin2 θ = + = d2 a2 c2 λ2 or sin2 θ =

λ2 4



l2 h2 + k2 + 2 2 a c

(4.3.4)

(4.3.5)

Similarly, we can show for an orthorhombic crystal (left as an exercise for the reader) that  λ2 h2 k2 l2 (4.3.6) sin2 θ = + + 4 a2 b2 c2 To summarize, the diffraction directions, 2θ , can be obtained purely from the size and shape of the unit cell. Example 4.3.1: Consider a powder diffraction experiment conducted on a polycrystalline specimen of cubic KI (density 3130 kg/m3 ) using Cu Kα ˚ radiation. The first eight lines were measured at the following (λ = 1.54 A) values of 2θ : 21.75, 25.19, 36.05, 42.45, 44.5, 51.7, 56.75, and 58.5. (a) Index these diffraction lines and calculate the d-values. (b) What is the lattice parameter, a0 ? (c) How many formula units, Z, are in the unit cell? ˚ r I− = (d) What structure do you expect for KI, given rK+ = 1.33 A, ˚ 2.20 A? Solution: (a) Applying Bragg’s law, (4.3.3), we get 4.08 Å (111), 3.53 Å (200), 2.5 Å (220), 2.13 Å (311), 2.04 Å (222), 1.76 Å (400), 1.621 Å (311), and 1.58 Å (420). ˚ we get the lattice parameter a0 = 4 d400 = (b) From d400 = 1.76 A, ˚ 7.04 A. (c) Comparing the experimental and theoretical density based on the measured lattice parameter, we get Z = 4. (d) For Z = 4, there are two possible structures, sphalerite or rock salt. But the radius ratio (∼0.61) suggests the NaCl (rock salt) structure (see Fig. 4.1.12).

OUP CORRECTED – FINAL

P k Incident Beam

ghkl

D Be iffra am ct e

d

Diffraction 259

2θ k0 O θ Ewald Sphere

2θ

Reciprocal Lattice Point

dhkl

Limiting Sphere

Figure 4.3.2 Interpreting diffraction in reciprocal space. The Ewald (reflecting) sphere construction for a set of planes, dhkl , at the exact Bragg angle; the vector, ghkl , satisfies the diffraction condition, k = k0 + ghkl . The limiting sphere of radius, 4π /λ, that includes all the possible reciprocal lattice points that can be diffracted is also shown. Note that the reciprocal lattice is fixed at the center, O.

4.3.2 The Ewald Construction: Interpreting Diffraction in Reciprocal Space We can also represent Bragg’s law in reciprocal space through the Ewald11 sphere construction. The incident, and diffracted beams are defined by their wave vectors, k0 and k, respectively, such that | k0 |=| k |= 2π/λ (Fig. 4.3.2). The origin of the reciprocal lattice is fixed at O, on the surface of the actual crystal. The Ewald, or reflecting, sphere, with radius, 2π /λ, lies within the limiting sphere of radius, 4π/λ, centered at O, and is positioned such that it just touches the limiting sphere at one end, and the origin, O, is at the other end, as shown. Note that k is parallel to the diffracted beam, making an angle, 2θ, with k0 , and intersects the Ewald sphere at a point, P. Bragg’s law is satisfied if, and only if OP is a reciprocal lattice vector, ghkl , of the crystal. Thus k = k0 + ghkl

(4.3.7)

where, ghkl , is perpendicular to the diffracting planes (hkl) with interplanar spacing, dhkl . This construction, attributed to Ewald, is the vector form of Bragg’s law (also known as the Laue criterion for diffraction) and can indeed be very useful. We can make three important conclusions of practical consequence from the Ewald construction: 1. The radius (4π /λ) of the limiting sphere is inversely related to the wavelength of the X-rays. If we reduce the wavelength, the radius of the limiting sphere increases, and larger reciprocal lattice vectors, or smaller inter-planar distances, are potentially included.

11 P. P. Ewald (1890–1985) was an English physicist.

OUP CORRECTED – FINAL

260

Crystallography and Diffraction 2. For the same incident beam direction, if the crystal is rotated about the origin, the reciprocal lattice of the crystal sweeps through the Ewald sphere. When a reciprocal lattice point intersects the Ewald sphere, Bragg’s law is satisfied, and diffraction with enhanced intensity is observed. We discuss the magnitudes of the diffracted intensity in §7.4. Further, even larger portions of the three-dimensional reciprocal lattice can be sampled by rotating the crystal about two orthogonal axes. We discuss experimental methods of X-ray diffraction in §7.9. 3. The reciprocal lattice points have a finite size determined by the physical size and perfection of the real crystals (§7.8; Fig. 8.5.4). Thus, diffraction can be observed even if the Ewald sphere does not pass exactly through a reciprocal lattice point, but in close proximity to it. This is discussed further in the chapter on electron diffraction (§8.5.3). These three rules assume purely coherent scattering without any loss in energy of the X-rays. Further, it is assumed that the incident beam is monochromatic and perfectly parallel. In practice, two principal deviations from these two conditions can be observed. Figure 4.3.3 shows the changes, using the Ewald sphere construction for (a) two different wavelengths of X-ray incidence, and (b) for poor collimation leading to beam divergence. In practice, Bragg’s law of diffraction is applied in two different ways. By using a crystal of known spacing, dhkl , and measuring the angle, θ , we can determine the wavelength, λ, of the X-ray radiation. This is the principle of

(a)

(b)

λ1 > λ2

Beam Divergence Ewald Sphere(s)

Limiting Sphere

Ewald Sphere(s)

Figure 4.3.3 The Ewald and limiting sphere construction illustrating (a) two different X-ray wavelengths, and (b) poor collimation and related beam divergence.

OUP CORRECTED – FINAL

Diffraction 261 wave-length dispersive spectroscopy as discussed in §2.5. Alternatively, it is used for crystal structure analysis where λ is known, θ is measured, and dhkl of various planes (hkl) in the crystal are to be determined. This is done by a number of diffractions methods using powder samples (Debye–Scherrer and diffractometer methods), single crystals (Laue and rotating-crystal methods), or thin films (X-ray reflectivity). We discuss these methods and other experimental details of X-ray diffraction in further detail in §7.9. However, in all cases, atomic positions within unit cells can only be established by measuring the diffracted intensities. In §7, we also develop a formulism to determine the diffraction intensities by first considering how X-rays are scattered by single electrons, then by an atom, and finally by the unit cell of the periodic crystal. Beyond §7 and §8, interested readers should consult Cullity (1978) and Hammond (2006) for further details of diffraction; for a more advanced perspective, see Woolfson (1997), Warren (1990), and Cowley (1975). Example 4.3.2: For the orthorhombic crystal in Example 4.2.1: (a) Apply Bragg’s law and calculate the angle, θ, of diffraction for all the planes using (i) Cu Kα, and (ii) Mo Kα, radiations. If you are interested in the (310) reflection, which radiation is better for this measurement? (b) Plot the two-dimensional real and reciprocal lattices (only a*-b* plane) for this crystal. (c) Draw the incident and diffracted vectors, and the Laue condition for diffraction of the (310) planes. Draw all angles accurately. Solution: ˚ and (a) The wavelengths for the two radiations are λCu, Kα = 1.54 A, ˚ λMo, Kα = 0.709 A. We apply Bragg’s law, (4.3.3), and solve it as a table: Plane (100) (002) (030) (011) (111) (310) (311) (312) 2.0 2.0 1.0 2.4 1.54 0.65 0.64 0.62 dhkl (Å) θ (Cu Kα ) 22.64◦ 22.64◦ 50.35◦ 18.71◦ 30.0◦ – – – θ (Mo Kα ) 10.2◦ 10.2◦ 20.76◦ 8.5◦ 13.31◦ 33.05◦ 33.63◦ 34.78◦ Based on the preceding table, it is not possible to use Cu Kα radiation to measure the (310); reflection and Mo Kα are preferred.

OUP CORRECTED – FINAL

262

Crystallography and Diffraction

(c) For θ = 33.05°

(b) k

ghkl 2θ

3Å 2Å

b

2π 3

Å–1

k0 b*

θ

π Å–1 dhkl

a a*

4.3.3 Comparison of X-Ray and Electron Diffraction The wave particle duality, corrected for relativistic effects, gives the wavelength of electrons accelerated through a voltage, V, in volts, as λ=

h = p

h

2me eV 1 +

eV me c 2



(4.3.8)

where me is the rest mass of the electron, e is its charge, and c is the speed of light. Typically, in a transmission electron microscope (§9), electrons are accelerated through a voltage in the range 100 kV < V < 400 kV, with corresponding wavelengths, λ, including relativistic corrections in the range 0.0037 nm > λ > 0.00164 nm. Note that, when relativistic effects are ignored, (4.3.8) can be simplified as h λ = = h(2me eV )−1/2 ∼ p



150 V

1/2 (4.3.8a)

where V is in volts and λ is in Å. Assuming 100 keV electron incidence, for diffraction from crystals with typical inter-planar spacing, dhkl ∼ 0.2 nm, the Bragg scattering angle, θ, is expected to be < 1◦ . At such small angles, sin θ = θ , and Bragg’s law of

OUP CORRECTED – FINAL

Diffraction 263 diffraction simplifies to λ = 2dθ . We can use this simplified form in the Ewald sphere construction for electron diffraction obtained in transmission electron microscopes (TEMs) using thin electron transparent specimens. The very short wavelength of electrons, compared to the inter-planar spacing, dhkl , in crystals, can modify the Laue equation, (4.3.7), i.e. |k0 | = |k| = 2π/λ ghkl , and affect the Ewald sphere construction (Fig. 4.3.4). The electron diffraction pattern, obtained in transmission, contains all the reciprocal lattice points intersected by the Ewald sphere. Because of the extremely shallow curvature, or very large radius of the Ewald sphere when compared to the reciprocal lattice, multiple regions

k0

k

Ewald Sphere

ZOLZ

ghkl FOLZ

ZOLZ

SOLZ FOLZ

Reciprocal Lattice Point

Figure 4.3.4 The Ewald sphere construction for electron diffraction, typically encountered in a transmission electron microscope. Because of the very short wavelength of the electrons the Ewald sphere radius is very large with a very flat surface, which intersects the reciprocal lattice in different zones and satisfying different zonal conditions. For a beam along [UVW], the lower image shows the zero order Laue zone, ZOLZ, with hU + kV + lZ = 0, and the ring of the first order Laue zone, FOLZ, with hU + kV + lZ = 1. The specimen is a single crystal of cubic spinel viewed along [001].

OUP CORRECTED – FINAL

264

Crystallography and Diffraction of diffraction spots, called Laue zones, can be simultaneously excited and seen in electron diffraction patterns (also see §8.6.3). First, a set of diffraction spots around the transmitted beam, 000, incident along a direction [UVW], called the zero order Laue zone (ZOLZ) is observed. The ZOLZ is so called because it also includes the origin, 000, in reciprocal space. Recall the zonal equation, (4.2.5), illustrated in Figure 4.1.10, and the related discussions. Thus, all the diffracted beams in the ZOLZ satisfy the zonal equation hU + kV + lZ = 0. In addition, because of its shallow curvature, the Ewald sphere can intersect reciprocal lattice points from planes with normal that are not parallel to the electron beam. These are called higher order Laue zones (HOLZ) and they are given specific names, i.e. first-order Laue zone (FOLZ) if they satisfy hU+kV+lZ = 1, second-order Laue zone (SOLZ), if they satisfy hU+kV+lZ = 2, and so on. Figure 4.3.4 includes an example of a diffraction pattern from a single crystal spinel (MgAl2 O4 ) specimen, illustrating the ZOLZ and FOLZ. It is instructive to compare typical scattering geometries for X-ray and transmission electron diffraction. We use an oriented single crystal thin film with an orthorhombic unit cell as an example (Fig. 4.3.5). For the θ -2θ X-ray scattering geometry shown (a), the measured lattice parameters would correspond to the family of d0k0 planes. On the other hand, if the same film were to be studied with a TEM, with plan view specimens, as shown in (b), the measured lattice parameters would be different and correspond to the dh00 family of planes. In other words,

(a)

(b)

k0

Diffracted

Incident

k Ewald Sphere

θ k g0k0

b* a*

b*

k0 Ewald Sphere

a* d0k0

gh00 dh00

Figure 4.3.5 (a) X-ray diffraction, including the Ewald sphere construction, for a typical θ-2θ scan of a single crystal thin film. Note that in this scattering geometry the inter-planar spacing of (0k0)-type planes are measured. (b) The same film measured by transmission electron diffraction. The scattering geometry, including the Ewald sphere, represents the diffraction condition shown and measures the (h00)-type planes.

OUP CORRECTED – FINAL

Diffraction 265 the two techniques measure inter-planar spacing of completely different families of planes, with significant consequence in the analysis of noncubic crystals. Electron diffraction is discussed further in §8 and §9, as well as in pedagogically excellent books by Williams and Carter (1996), Cowley (1984), and Spence and Zuo (1992). Here, from this brief introduction, the important point to take away is that in typical transmission electron diffraction the wavelength of the electrons is much smaller than the inter-planar spacing in crystals, i.e. λe 10 eV, has a wavelength much less than the inter-atomic spacing, and can easily penetrate a solid. Further, the de Broglie wavelength, (1.3.3), is inversely proportional to the square root of the atomic mass; thus, heavier atoms and the nuclei of such ions can more easily penetrate a solid. However, in ion–solid interaction, the important parameter is not

OUP CORRECTED – FINAL

Interactions of Probes with Matter, Including Damage 305

Cross-section (b)

103 102

PIX

E Lα

2 MeV H+ PIXE Kα

10¯11

10 BS

2 MeV He R

1

10¯12

0.1 10¯2 10¯3

NRA

Effective interaction diameter (cm)

10¯10

2M eV H +

10¯13 20

40 60 Atomic number, Z

80

Figure 5.3.12 Cross-sections in units of barns (1b = 10−24 cm2 /atom) for the interactions of protons and Helium ions as a function of atomic number. Scattered data points for nuclear reaction analysis (NRA) are also shown. Adapted from Bird and Williams (1989).

the distance the ion travels in a solid, but the number of atoms the ion encounters, as defined by the atomic areal density (number of atoms/cm2 ), ρ A . In addition, the probability of a particular interaction between the ion and the atom is given by the cross-section, σ , defined as the effective area of the atom or nucleus seen by the ion. Thus, the probability, P, that a projectile ion will have a specific interaction is given by the product, i.e. P = ρA σ . Figure 5.3.12 shows cross-sections for the interactions of high-energy ions and atoms, as a function of atomic number, relevant for materials analysis. The average distance between collisions is defined as the mean free path length (§1.3.5.1), and is inversely proportional to the cross section. If the cross-section is high, the mean free path length is low, and the ion beam is significantly attenuated as it propagates in the solid. The attenuated beam flux is given by the same equation that we used to describe X-ray absorption (2.4.2), but with the parameter, μ, representing the attenuation coefficient for the ion beam. Note that the same equation, (2.4.2), applies to all cases of photons, ions, and neutrons, that are attenuated in a solid without undergoing any energy loss. 5.3.5.2

Ion–Solid Interactions

Figure 5.3.13 schematically illustrates the interaction of ions with a specimen. Broadly speaking, it is possible to divide the ion–specimen interaction into two categories. Kinematic collisions between the primary ion (mass, M 1 , and initial energy, E 0 ) and the atoms (mass, M 2 , and initial energy, E = 0, i.e. at rest) of the specimen, leading to their scattering by angles, θ 1 and θ 2 , with energies, E 1 and E 2 ,

OUP CORRECTED – FINAL

306 Probes: Sources and Their Interactions with Matter ion cludters +

(b) Higher energy electrons ions penetrate few atomic layers, and ejects electrons, atoms and ions (SIMS)

(c) Very high energy light ions penetrate much deeper and backscattered at different depths, and continuously lose energy (RBS)

– (a) Low energy ions are reflected (LEISS)

+ –

ions atoms

Thousands of atomic layers

Figure 5.3.13 Ion–solid interactions: (a) Low energy primary ions are reflected, and form the basis of low-energy ion-scattering spectroscopy (b) Higher energy ions penetrate, at most, a few atomic layers, undergoing many major interactions and causing collision cascades. In the process, electrons, atoms and ions ejected from the solid, can be analyzed in a technique known as secondary ion mass spectrometry (c) Very high-energy light ions penetrate many thousands of atomic layers, occasionally undergoing major interactions and being back-scattered; however, they also suffer inelastic scattering and continuously lose energy as they propagate in the solid. Analysis of the energy distribution of the back-scattered ions forms the basis of Rutherford back-scattering spectroscopy (RBS).

respectively (Fig. 5.3.15). In this process, very high-energy (MeV) primary ions are back-scattered, even from substantial depths in the solid. In addition, both incoming and scattered ions continuously lose energy by inelastic scattering processes, which has to be accounted for in the scattering kinematics (Fig. 5.3.17; Fig. 5.4.2). Alternatively, when higher-energy ions strike a specimen, they penetrate it, and following a number of collisions with atoms are slowed down and brought to rest. Now, the kinetic energy of the primary ion is distributed over several atoms, which in turn, strike other atoms, causing a collision cascade. As a result, the primary ion travels in an irregular path before coming to rest in the specimen (Fig. 5.3.13b).

OUP CORRECTED – FINAL

Interactions of Probes with Matter, Including Damage 307

Rp

Figure 5.3.14 Schematic drawing defining the projected range, Rp , depth of penetration, dp , and the path, l, of an ion in a solid, and defining its range, R, which is a straight line from the point of entry to its final resting point.

l R ψ O

dp

This process is also known as ion implantation. The distance from the point of entry traveled by the ion in the solid before it comes to rest is called its range, R, and its projection along the direction of incidence is called the projected range, Rp (Fig. 5.3.14). Note that the actual path, l, travelled may be larger than R. For normal incidence, ψ = 0◦ , the projected range is the same as the penetration depth, i.e. dp = Rp . The atoms in the specimen struck by the collision cascade are also displaced, more or less in an isotropic angular distribution from the original sites. Atoms on the surface of the specimen may acquire sufficient energy in the collision cascade to overcome the surface binding forces and be ejected from the specimen. This process is called sputtering, and it is sometimes used as a way to remove surface layers, atomic layer by atomic layer. Further, after a layer of the specimen is removed by sputtering, the specimen can be chemically analyzed using a surface analytical technique, such as Auger spectroscopy, to give chemical information as a function of depth (Fig. 2.6.9). Last, but not least, secondary ions, and clusters of ions and atoms are also ejected from the specimen surface (Fig. 5.3.13). These secondary ions can be detected and their mass analyzed by a technique called secondary ion mass spectrometry or SIMS (§5.4.3), a destructive method that is nevertheless used routinely in surface analysis. 5.3.5.3

Physics of Kinematic Collisions

The scattering between an energetic primary ion and the atoms in a specimen can be described as a classical two-body elastic collision, interacting through a centrosymmetric potential. Figure 5.3.15 shows the trajectories of the two masses, M 1 and M 2 , before and after collision, in the laboratory frame. M 1 is the projectile/primary ion with energy E 0 , and the target atom, M 2 , is initially at rest, i.e. E = 0. After collisions, their energies are E 1 and E 2 , and the scattering angles are θ 1 and θ 2 , respectively. The distance of closest approach, Dc , for a head-on collision, D0 , defined for a zero-impact parameter (Fig. 5.3.15) is important to determine the nature of the ion-atom collision. It is given by D0 = 1.44 × 10−9 (Z1 Z2 ) (M1 + M2 ) / (M2 E0 )

(5.3.9)

OUP CORRECTED – FINAL

308 Probes: Sources and Their Interactions with Matter Scattered Ion, M1, E1, v1

y

Figure 5.3.15 Kinematic scattering geometry between a primary/projectile ion and a stationary atom in the specimen. The impact parameter, D1 , which is the separation between the ion and atom (center to center) perpendicular to the initial trajectory of the ion, and the distance, Dc , of closest approach that determines the nature of the ionatom interaction is shown. Typical value for keV ions (LEISS, §5.4.2) with λ ∼ 10–3 nm is Dc ∼ 0.05 nm, and for MeV ions (RBS, §5.4.1) with λ ∼ 10–5 nm, is Dc ∼ 0.001 nm.

Primary Ion, M1, E0, v

x θ1

Dc

D1 Electron Cloud Nucleus

θ2 Target Atom M2, E=0

Recoil Atom M2, E2, v2

where E 0 is in eV, M is the atomic mass in amu, Z is the atomic number, and D0 is in meters (Fig. 5.3.15). If D0 is less than the sum of the atomic or ionic radii, excitations can take place, and the classical kinematic collision described here must be replaced by a quantum mechanical description. Example 5.3.1: Are the collisions with a K+ ion (radius ∼0.138 nm) for a He+ ion (atomic radius ∼0.031 nm) accelerated at (1) 2 MeV (RBS) and (2) 2 keV (LEISS), kinematic? Solution: We apply (5.3.9) with Z1 = 2, M1 = 4, Z2 = 19, M2 = 39.1 and determine D0 . (1) For 2 MeV He+ ion, D0 = 3.0 × 10−14 m = 3 × 10−5 nm, which is smaller than the sum of the two diameters. (2) For 2 keV He+ ions, D0 = 3.0 × 10−11 m = 3 × 10−2 nm, which is also smaller than the sum of the two diameters. In both cases, it will cause excitation (or inelastic scattering, where the incident ion loses energy) and a simple kinematical collision model cannot be applied. Nevertheless, we shall do so, but assume that the inelastic scattering gives rise to a continuous back-scattered energy spectrum up to a maximum energy dictated by the kinematic collision for the heaviest atomic mass in the specimen (see RBS spectrum, Fig. 5.3.17).

OUP CORRECTED – FINAL

Interactions of Probes with Matter, Including Damage 309 We define the dimensionless, kinematic factor, K1 = E1 /E0 = v21 /v2 , for the projectile/primary ion. From a conservation of energy, we get 1 1 1 M1 v2 = M1 v21 + M2 v22 2 2 2

(5.3.10a)

Now, conserving the momentum along the direction of motion (x) of the primary ion we get M1 v = M1 v1 cos θ1 + M2 v2 cos θ2

(5.3.10b)

and similarly, for conservation of momentum perpendicular to the direction of motion ( y) gives 0 = M1 v1 sin θ1 − M2 v2 sin θ2

(5.3.10c)

Eliminating θ 2 first, and then v2 , from (5.3.10 a–c) we get v1 = v

1/2  + M1 cos θ1 ± M22 − M12 sin2 θ1 M1 + M2

(5.3.10d)

from which we can determine the dependence of K 1 as a function of the scattering angle, θ 1 , as ⎛ 1/2 ⎞2  cos θ1 ± A2 − sin2 θ1 E1 ⎜ ⎟ K1 (θ1 ) = =⎝ ⎠ E0 A+1

(5.3.11)

Note that the kinematic factor, K 1 , only depends on the mass ratio, A = M2 /M1 , and the scattering angle, θ 1 . The positive sign holds for A > 1, and both signs for A < 1. The corresponding equation for the recoiling target atom is K2 (θ2 ) =

E2 4A = cos2 θ2 E0 (1 + A)2

(5.3.12)

with a maximum value when θ2 = 0◦ . Figure 5.3.16 plots the function, K1 (θ1 ), based on (5.3.11), for different values of A. Two specific cases are of interest in materials analysis. For θ1 = 90◦ , K1 (θ1 ) is particularly simple, i.e.   E1 A−1 M2 − M1 = = K1 90◦ = E0 A+1 M2 + M1

(5.3.13)

OUP CORRECTED – FINAL

310 Probes: Sources and Their Interactions with Matter 1 100 0.8

20 10

0.6

0.2

K1 0.4 Figure 5.3.16 The kinematic factor, K 1 , as a function of the scattering angle, θ 1 , in the laboratory frame of reference. The parameter, A = M2 /M1 , is the ratio of the scattering atomic mass (M 2 ) to the primary/ projectile ion mass (M 1 ).

5

3 0.2

0.4

2 A=1

0.8

0 0°

30°

60°

90°

120°

150°

180°

θ1

Adapted from Bird and Williams (1989).

which can be rearranged as  M2 = −M1

E1 + E0 E1 − E0

 (5.3.14)

Thus, if M 2 , or the atomic mass of the element(s) in the specimen, is unknown, it can be determined by measuring the energy of the projectile after collision. Similarly, for θ1 = 180◦   K1 180◦ =



A−1 A+1



2 =

M2 − M1 M2 + M1

2 =

E1 E0

(5.3.15)

The relationships (5.3.11) and (5.3.12) can be used to identify the scattering atoms from ion energy spectra. In fact, they are sufficiently sensitive to be able to resolve individual isotopes of light elements with MeV 4 He+ ions. Figure 5.3.17 shows typical energy spectra for scattering of 4 He+ ions at θ1 = 140◦ , from a specimen containing 108 Ag, 28 Si, and 16 O, for two different energies, E0 = 1 keV (LEISS, §5.4.2) and E0 = 1 MeV (RBS, §5.4.1). Example 5.3.2: For the two cases discussed in Example 5.3.1, calculate the kinematic factor assuming that the scattering angle is 170◦ . Solution: We use (5.3.11), with A = M2 /M1 = 39.1/4 = 9.785, Cos θ1 = −0.985, Sin θ1 = 0.174. Then

OUP CORRECTED – FINAL

Interactions of Probes with Matter, Including Damage 311 Si

108Ag 4

He

28Si 16O

Scattered ion yield

θ = 140°

Scattered particle yield

LEISS

Figure 5.3.17 Schematic representation of the scattering of 4 He+ ions, at an angle of 140◦ , from bulk silicon with atoms of Ag, Si and O, on its surface. Two representative spectra for low-energy, E0 = 1 keV (LEISS, upper), and high-energy, E0 = 1 MeV (RBS, lower) ions are shown.

RBS

0

0.2

0.4 0.6 E1/E0

0.8

1.0

Adapted from Vickerman and Gilmore (2009).

1/2

(1) E0 = 2 MeV, and K1 = [{−0.985 + (9.7852 − 0.03) 1)]2 = 0.6655 Note that this agrees with Figure 5.3.16.

}/(9.785 +

(2) For E0 = 2 keV, K1 = 0.6655. Clearly, the kinematic factor does not depend on E 0 . For light ions, such as 4 He+ , penetrating a solid, both the incident and scattered ions lose energy primarily through excitations and ionizations of atomic electrons. These inelastic collisions are microscopically discrete, but macroscopically, we assume that the moving ions lose energy continuously. Thus, any inelastic scattering that occurs at some depth below the surface, gives rise to a continuous backscattered energy spectrum up to a maximum energy dictated by the kinematic collision for the heaviest atomic mass in the specimen. This can be seen clearly for silicon in the RBS spectrum (Fig. 5.3.17). Further details of LEISS and RBS, and the interpretation of the respective spectra, are discussed in §5.4.

OUP CORRECTED – FINAL

312 Probes: Sources and Their Interactions with Matter 5.3.5.4

Specimen Damage

If primary ions with sufficient kinetic energy strike a specimen, apart from the small fraction ( ∼1%) that is reflected from the surface, they generally penetrate it and are slowed down by elastic (nucleus) and inelastic (electrons) collisions. The former results in a collision cascade with a number of possible consequences: 1. The primary ion will traverse a path, l, and range, R (Fig. 5.3.14) and will eventually come to a stop inside the specimen. This is referred to as ion implantation and can lead to a local change in composition at some depth in the interior of the specimen. In the semiconductor industry such ion implantation is used to dope wafers at specific depth locations. 2. The collision cascade displaces the atoms in the solid from their original crystallographic sites. The resultant change in atomic configuration can lead to a complete loss of long-range order (amorphization) and/or atomic mixing, including the implanted primary ions, in the specimen. 3. A very small number of atoms in the specimen may experience direct impact from the primary ions. Such atoms experience an energy transfer larger than in the cascade and are pushed in the forward direction. This leads to so-called recoil mixing, and results in local changes, most significantly, in buried interfaces. Similarly, surface contaminants and adsorbates, if present, can also experience a direct impact and be pushed forward into the interior of the specimen; this process is called recoil implantation. 4. If the specimen is highly insulating, it can become charged upon primary ion irradiation. Then, an electric field extending far into the interior of the specimen can be formed, which can promote the motion of charged ions. This process, known as electromigration, has been observed for alkali ions in SiO2 . 5. The atoms on the surface (uppermost monolayers) of the specimen, stuck by the collision cascade may possess sufficient kinetic energy to escape the solid by overcoming the surface binding forces. This process is known as ion-beam sputtering (see §5.3.5.5). These processes, are covered in detail in more specialized books on ionmaterials interactions (see Bird and Williams, 1989). From a materials analysis point of view, ion-beam sputtering (5), is of particular interest as it forms the basis of depth profiling (systematic removal of surface layers, one monolayer or so at a time, followed by chemical analysis, such as Auger electron spectroscopy). Alternatively, the sputtered ions are directly mass-analyzed, as in SIMS (§5.4.3), to provide chemical information; in this case, the yield of secondary ions is affected by (1–4), which are collectively known as matrix effects.

OUP CORRECTED – FINAL

Interactions of Probes with Matter, Including Damage 313 5.3.5.5

Ion-Beam Sputtering

The rate of sputtering, dNi /dt, for atoms of element, i, in the collision cascade model (Fig. 5.3.13) defined as the number of sputtered atoms per unit time, is proportional to the primary ion flow rate, dN0 /dt = I0 /q and given by dNi /dt = Yi dN0 /dt

(5.3.16)

where I 0 is the current of ions with charge, q, and the proportionality factor, Yi , is called the sputtering yield. In the case of a single element specimen, the yield, Yei (E0 ), will depend on  how much energy the primary ion will deposit close to the surface, (dE/dz)n z=0 , and the binding strength of the surface atoms. Thus, for a specific ion–specimen combination, we have  Yei (E0 ) ∼ (dE/dz)n z=0 /Esub

(5.3.17)

where the numerator is the specific loss, due to collisions with the nuclei on the surface, and the binding energy of the surface atoms is approximated by the sublimation energy, E sub , of the material. For example, Figure 5.3.18 shows the values for the sputtering yield of Al, as a function of the primary ion energy, E 0 . While the sputtering yield will depend on the crystallinity and topography of the specimen, it can be seen that the yield (Fig. 5.3.18) is a maximum for E 0 ∼ 10 keV; a typical value often used in practice for sputtering. Moreover, for fixed E 0 , the sputtering yield will vary by a factor of 3–5 across the periodic table. For a multi-element specimen, (5.3.17) is modified to Ysi = Xis Yei

(5.3.18)

102

Yield

101 100 Ar(expt) 10–1

He (expt) Xe(expt)

10–2

Xe(theory) Ar(theory)

10–3 0.01

0.1

1

10 100 Energy (keV)

1000

10000

Figure 5.3.18 Sputter yield data for aluminum using different primary ions as a function of ion energy. Notice the peak at ∼ 10 keV. Adapted from Vickerman and Gilmore (2009).

OUP CORRECTED – FINAL

314 Probes: Sources and Their Interactions with Matter where, Xis is the fractional surface concentration of element, i, in the specimen. In practice, due to the matrix effects, (1–4), discussed earlier in §5.3.5.4, the specimen composition is affected by the ion–specimen interactions, and the sputtering yield is altered from Yei to Ysi . Thus, for a multi-element specimen, the total sputtering yield, Y, is Y = Xis Ysi =

dN/dt dN0 /dt

(5.3.19)

Further, the number of ions produced by sputtering is given by the secondary ion yield ± s Y± i = Pi Yi

(5.3.20)

where, Pi± is the ionization probability. In practice, the instrument detection parameters, βi , are included and the secondary ion intensity is written as Ii± = βi Pi± Ysi (dN0 /dt)

(5.3.21)

Since the ionization process is affected by the electronic state of the surface, the secondary ion yields vary by several orders of magnitude across the periodic table (Fig. 5.3.19). They are also dependent on the chemical state of the surface; thus, we can expect the ion yields to be different for a metal and its oxide.

Ca Mg 6 Be Al B

Ti Ru In Ba V Cr Mo Ga Mn Zr Si Fe Nb

Elements Re W Hf

Co

5

Compounds

Pd Ni

Log I+

Os Ta Th

Ge

P

Sn

Zn

4 H

Ir

Pb

Ag Sb

Figure 5.3.19 Positive ion yield as a function of atomic number for elements (open circles) and compounds (filled squares). A primary current of 1 nA, 13.5 keV, O-ions were used. Adapted from [8].

3

2 N 0

C O

As Sb Cd

Bi

Te

Hg Pt

S

Au 20

40 60 Atomic number, Z

80

100

OUP CORRECTED – FINAL

Ion-Based Characterization Methods 315

5.4 Ion-Based Characterization Methods A high-energy (MeV) primary ion penetrating a solid will undergo energy transfer by a variety of collision processes (Fig. 5.3.13). The most common process is inelastic scattering with the valence electrons, and involves energy losses of the order of 10s of eV per collision, but with insignificant deviation in the trajectory (i.e. no deflection) of the primary ion beam. The cross-section for such processes is high and of the order of 10–16 cm2 , with correspondingly large impact parameters of ∼1 Å, i.e. of the order of typical lattice parameters in materials. At smaller impact parameters, collisions leading to inner-shell ionization and subsequent deexcitation by the emission of characteristic X-rays are observed. Such chemical analysis of materials by PIXE is discussed in §5.4.4. If such a collision leads to back-scattering (θ 1 ∼ 140◦ –180◦ ; Fig. 5.3.15), then the final energy of the scattered ion will depend on the elastic nuclear collision at a given depth combined with a number of inelastic collisions, each with a small energy-loss, as the primary ion travels into and the back-scattered ion propagates out of the specimen. Thus, measuring the energy of the scattered ion not only gives the mass of the scattering atom in the specimen but also its depth location within the specimen. This is the basis of RBS, discussed in §5.4.1. It is worth emphasizing here that the kinematic collision and the associated scattering cross-section are independent of the chemical bond, and hence the back-scattered spectrum is independent of the nature of the bonding or the electronic structure of the specimen. Compared to RBS, which yields an edge feature (Fig. 5.4.3) in the spectrum, LEISS, as in Figure 5.3.17, is characterized by sharp peaks for each atomic species found on the specimen surface. In LEISS, only ions back-scattered from the surface that retain their original charge, i.e. not neutralized, are detected with appropriate electrostatic analyzers. LEISS is discussed in §5.4.2. Last, but not least, the secondary ions produced upon primary ion bombardment can be detected and their mass analyzed; this technique of SIMS is discussed in §5.4.3.

5.4.1 Rutherford Back-Scattering Spectroscopy (RBS) One parameter of importance for RBS is the energy loss of the primary ions per unit length—dE/dz—due to the inelastic scattering in the specimen. Commonly it is expressed in terms of the stopping power, S (eV/(atoms/cm2 )) where S=−

dE 1 dz N

(5.4.1)

and N is the atom density (number of atoms/cm3 ). The stopping power data for all elements has been tabulated (Zeigler, 1977), and (5.4.1) is used extensively in RBS analysis to calculate the energy loss per unit length. Figure 5.4.1 shows a typical plot of the stopping power for 4 He+ in Ni. The broad peak at ∼1 MeV is normally used as the operating energy for RBS.

OUP CORRECTED – FINAL

Figure 5.4.1 Stopping power of 4 He+ ions in Ni as a function of incident ion energy. Notice the broad peak at ∼1 MeV.

Stopping power (eVcm2/1015 atoms)

316 Probes: Sources and Their Interactions with Matter 80

4

He

Ni

60

40

20

0 10

1

Adapted from Vickerman and Gilmore (2009).

103 100 Incident ion energy (keV)

104

105

For compounds and mixtures, a simple additive rule of energy loss may be used. Thus, the stopping power for a compound, Am Bn , is given by SAm Bn = mSA + nSB

(5.4.1a)

where m + n is normalized to unity. The stopping power for 4 He and 1 H are tabulated for select elements and energies in Table 5.4.1.

Example 5.4.1: Calculate the stopping power for Si3 N4 for 3 MeV 4 He ions. Solution: Table 5.4.1 gives S Si (3 MeV) = 39.17 eV cm2 /1015 atoms, and S N (3 MeV) = 25.01 eV cm2/ 1015 atoms. From (5.4.1a), we get SAm Bn = (3 × 39.17 + 4 × 25.01) /7 = 31.08 eV cm2/ 1015 atoms.

Table 5.4.1 Proton (1 H) and Helium (4 He) stopping power (eVcm2 /1015 atoms) for select elements and energies. 1H

4 He

Z Element 100 keV 1 MeV 2 MeV 3 MeV 4 MeV 100 keV 1 MeV 2 MeV 3 MeV 4 MeV 1 2 3 4 5

H He Li Be B

4.690 6.123 8.120 10.044 13.653

1.125 1.906 2.662 3.296 4.045

0.656 1.116 1.638 2.064 2.488

0.474 0.808 1.208 1.530 1.834

0.374 0.638 0.965 1.226 1.465

7.66 9.14 13.51 22.92 23.14

11.87 17.42 22.02 24.24 33.75

7.68 12.48 16.29 19.15 24.58

5.62 9.43 12.82 15.63 19.44

4.5 7.64 10.67 13.22 16.22

OUP CORRECTED – FINAL

Ion-Based Characterization Methods 317 Table 5.4.1 Continued 1H

4 He

Z

Element 100 keV 1 MeV 2 MeV 3 MeV 4 MeV 100 keV 1 MeV 2 MeV 3 MeV 4 MeV

6 7 8 9 11 12 13 14 15 16 17 19 20 22 23 24 25 26 27 28 29 30 31 32 33 34 39 41 42 46 47 48 49 50 56 57 78 79 82

C N O F Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Y Nb Mo Pd Ag Cd In Sn Ba La Pt Au Pb

14.428 16.222 16.145 13.434 20.975 20.700 20.460 26.428 27.108 27.706 30.424 32.604 27.434 30.923 32.362 29.221 27.737 29.619 27.314 25.010 23.737 24.168 25.262 27.976 32.061 31.135 41.148 42.281 28.632 36.105 34.727 39.929 38.637 40.480 56.156 54.859 36.452 35.401 44.030

4.602 5.154 5.737 6.170 7.680 7.605 7.786 8.271 8.773 8.182 9.596 11.067 11.140 11.771 12.043 12.065 12.153 12.358 12.358 12.557 12.616 12.869 12.848 13.602 13.326 13.657 15.751 17.371 16.170 17.511 16.982 17.530 17.837 17.627 20.675 20.686 20.522 20.968 21.859

2.866 3.209 3.602 4.740 4.826 4.854 4.994 5.288 5.634 5.313 6.219 6.879 7.197 7.528 7.691 7.767 7.864 8.039 8.105 8.366 8.524 8.657 8.798 9.413 9.145 9.420 10.694 11.361 10.673 11.714 11.471 11.754 11.972 11.817 13.817 13.512 14.915 15.094 15.664

2.131 1.714 2.404 1.943 2.700 2.183 3.343 2.934 3.600 2.904 3.646 2.949 3.760 3.049 4.002 3.264 4.270 3.482 4.081 3.355 4.773 3.922 5.159 4.193 5.520 4.537 5.687 4.627 5.865 4.812 5.911 4.845 6.000 4.926 6.156 5.066 6.231 5.141 6.456 5.335 6.593 5.445 6.729 5.589 6.886 5.728 7.377 6.137 7.172 5.977 7.387 6.157 8.386 6.992 8.741 7.223 8.256 6.847 9.096 7.558 8.998 7.524 9.153 7.623 9.323 7.762 9.212 7.676 10.740 8.926 10.446 8.680 15.040 10.234 12.217 10.414 12.667 10.784

Note: values for intermediate energies can be linearly interpolated. Adapted from Tesmer and Nastasi (1995).

23.62 26.91 26.70 22.80 40.46 44.33 40.64 41.84 45.21 45.49 53.23 47.14 50.77 49.40 45.77 43.02 40.06 43.68 41.33 39.97 34.95 33.71 44.73 51.38 62.62 49.33 73.67 76.22 55.02 68.64 56.37 57.80 56.85 64.87 87.15 79.39 49.79 53.96 72.95

37.78 45.21 46.17 43.28 54.99 55.61 53.18 63.71 70.21 73.02 85.27 85.90 82.55 86.81 96.76 83.70 82.76 85.12 82.50 76.11 75.15 77.22 81.60 83.05 88.62 87.64 116.25 119.39 109.86 109.38 111.17 114.85 116.53 119.46 147.60 148.21 117.60 122.53 139.14

27.83 32.54 34.92 36.34 44.49 44.11 43.99 48.33 51.56 50.69 59.38 66.59 66.4 68.49 73.21 69.18 69.20 70.29 69.48 68.03 67.14 69.50 70.18 72.06 72.76 73.06 90.23 98.62 91.09 95.79 94.98 96.66 98.45 98.61 116.41 119.16 104.40 110.71 118.03

22.05 25.01 27.54 29.41 36.41 35.95 36.56 39.17 41.53 39.20 45.98 53.15 53.13 55.71 57.72 56.93 57.20 58.09 57.85 58.08 57.89 59.41 59.22 62.06 61.35 62.48 73.56 81.60 75.70 81.22 79.14 81.40 82.82 82.14 98.48 97.64 92.07 95.39 99.98

18.46 20.67 23.01 24.75 30.81 30.52 31.25 33.20 35.21 32.84 38.52 44.41 44.72 47.27 48.35 48.46 48.82 49.65 49.65 50.47 50.72 51.73 51.66 54.70 53.59 54.93 63.34 69.85 65.03 70.46 68.33 70.54 71.78 70.94 83.23 83.25 82.81 84.59 88.19

OUP CORRECTED – FINAL

318 Probes: Sources and Their Interactions with Matter Example 5.4.2: A binary compound, Nax Cl1-x , with a density of 8 × 1022 atoms/cm3 , is analyzed by RBS, using 2 MeV 4 He ions. The ions lose energy at the rate of 4.16 × 1039 eV/cm. (a) Calculate the stopping power for this compound. (b) What is the composition of the binary alloy? Solution: (a) The stopping power, (5.4.1), for the compound can be calculated as     SBC = 4.16 × 1039 / 8 × 1022 = 5.19 × 1016 eVcm2 /atom = 51.9 eV cm2 /1015 atoms. (b) From Table 5.4.1, the stopping power for the two elements are SNa = 44.49 eV cm2 /1015 atoms, and SCl = 59.38 eV cm2 /1015 atoms From (5.4.1a), we have SBC = x SNa + (1 − x) SCl , which gives 51.9 = x 44.49 + (1 − x) 59.38. Thus, x = 0.5, and the compound is Na0.5 Cl0.5 or NaCl.

5.4.1.1

Energy Width in Back-Scattering Spectroscopy

We now determine the energy width in the RBS spectrum corresponding to the finite layer thickness of a specimen and illustrate it with a simple example (Fig. 5.4.2). The total energy loss of the primary ion with energy, E 0 , as it penetrates a thin film of thickness, t, is Ein = 0



where,

dE  dz in

t

 dE  dE dz = t dz dz in

(5.4.2)

  is an average value between E 0 and E0 − t dE dz  . Then, at any

depth, t, the energy of the ion in the specimen is E(t) = E0 − t

 dE  dz in

in

(5.4.3)

After large angle (θ 1 ) back-scattering, the energy of the ion just after backscattering is K 1 E(t), where K 1 is the kinematic factor, (5.3.11). In the outward

OUP CORRECTED – FINAL

Ion-Based Characterization Methods 319 (a)

(b)

K1Au(170o)=0.923

10

2.91 MeV

Yield (a.u.)

3 MeV 4He+

Al

8 AuB

6

AuF

2.69 MeV

AuB AuF

4000 Å ΔEAl ∼0.165 MeV

4

3 MeV

dE dz (c)

2.77 MeV

(3 MeV) = –22 eV/Å 2.59 MeV

Al

2.91 MeV

3 MeV

Al 2

1.6 MeV

0

dE (1.6 MeV) = –29 eV/Å dz Al

1.5 1.65

2.0 Energy (MeV)

2.5 2.59 2.77

1.65 MeV 1.484 MeV

Al

K1 (170o) = 0.55

Figure 5.4.2 (a) Back-scattering spectra of 3 MeV 4 He+ ions of a 400 nm thick Al film with two Au surface markers. (b) For the AuF layer one sees a sharp peak at 2.77 MeV; the incident ion losses energy (0.088 MeV) traversing the Al film and arrives at the AuB layer with an energy of 2.91 MeV. Upon back-scattering it has an energy of ∼ 2.69 MeV, loses energy ( ∼ 0.1 MeV) in the Al film, and emerges with an energy of 2.59 MeV, appearing as a sharp peak. (c) On encountering the front surface of the Al layer, the He ion is back-scattered at an energy of 1.65 MeV. It loses energy in the Al layer and arrives at the back surface with an energy ∼ 2.91 MeV. Upon back-scattering at the very back layer it has an energy of ∼ 1.6 MeV, loses energy (0.116 MeV) in the Al film, and emerges with an energy of 1.484 MeV. Note that as the ion traverses into the film it is back-scattered at various depths and as a result, the back-scattered yield for Al has a broad shaped peak with a width of 0.165 MeV, as shown. Adapted from Feldman and Mayer (1986).

(back-scattered) path, the ion continues to lose energy and its energy, Eout (t) at any depth, t, is Eout (t) = K1 E(t) −

     t 1 dE  dE  dE  = −t K + + K1 E0 1 | cos θ1 | dz out dz in | cos θ1 | dz out (5.4.4)

Thus, for a film of thickness, t, the energy width, E, in the RBS signal is     1 dE  dE  + E = t K1 = t [S] dz in | cos θ1 | dz out

(5.4.5)

OUP CORRECTED – FINAL

320 Probes: Sources and Their Interactions with Matter where, [S] is the back-scattering energy-loss factor. In the surface energy approximation, for typical films with t ≤ 100 nm, assuming that the change in energy along the path of the ion is negligible, (5.4.5) simplifies to    dE  dE  1 = t [S0 ] E0 = t K1 + dz E0 | cos θ1 | dz K1 E0 

(5.4.6)

The implications of (5.4.6) are illustrated in Figure 5.4.2a, which shows an RBS spectrum, using 3 MeV 4 He+ ions, for a standard test film of Al, 400 nm thick, with Au markers on the top and bottom surface. Features of the RBS spectrum, for the Au markers (Fig. 5.4.2b) and the Al layer (Fig. 5.4.2c)   are also explained. For Al, dE ∼ 22 eV/Å for 3 MeV ions (stopping power,  dz in  15  22 2 3 S = 36.56 eV/ 10  atoms/cm ) and NAl = 6×10 atoms/cm ) back-scattered  ∼ 29 eV for ∼1.5 MeV (S = 48.34 eV/(1015 atoms/cm2 )), at 170◦ , and dE dz  out

with K 1 (Al) ∼ 0.55; note that the energy loss rate is determined from known values of stopping power, S, as in (5.4.1) and known values of the atom density, N. Substituting these values in (5.4.5), we get E (Al) ∼ 165 keV for the film. This is very close to the observed separation (∼175 keV) of Au peaks, but the difference is understandable as K 1 (Au) is used in (5.4.5). 5.4.1.2

Shape of the Back-Scattering Spectrum

Is there a well-defined shape to an ideal spectrum in RBS data? To answer this question, we consider a back-scattering experiment from an infinitely thick Au target, a detector with a solid angle, , and N 0 incident 4 He+ particles of 1.4 MeV energy (Fig. 5.4.3). Then the yield, Y(t), from a thin layer of thickness, t, with Nt target atoms/cm2 in the layer is Y = σ (θ)  N0 Nt

(5.4.7)

where σ (θ ) is the scattering cross-section for elements of atomic numbers, Z 1 (primary particle) and Z 2 (matrix atom), given by  σ (θ ) 

Z1 Z2 e2 4E1 (t)

2 (5.4.8)

and E 1 (t) is the energy of the particle at depth, t. The ratio, A, of the energy lost in the outward path, Eout = K1 E(t) − E1 , and the inward path, Ein = E0 − E(t), where E 1 (t) is the measured energy, is given by A=

Eout K1 E(t) − E1 (dE/dz)out = = Ein E0 − E(t) (dE/dz)in

(5.4.9)

OUP CORRECTED – FINAL

Ion-Based Characterization Methods 321

Au

E0 = 1.4 MeV 4He E1(t)

t

8

Yield (counts × 10–3)

t 6

Experiment

4

Y(t) Figure 5.4.3 The back-scattering yield for 1.4 MeV 4 He+ ions incident on a thick gold specimen. The dashed line is calculated using (5.4.11) and normalized to the experiment.

2

E1(t) 0 0.6

0.8

1.0 1.2 Energy (MeV)

1.4

Adapted from Feldman and Mayer (1986).

For MeV electrons, A is a constant, and then by rearranging terms we get E(t) =

E1 + AE0 K1 + A

(5.4.10)

For medium to heavy mass target atoms, K1 = 1 and A = 1, and thus  Y (E1 ) ∝

1 E0 + E1

2 (5.4.11)

which accounts for the typical spectral shape in RBS spectra (Fig. 5.4.3). 5.4.1.3

Ni Thin Films Grown on Silicon: A Technological Example

A routine problem in silicon semiconductor films is the formation of metal silicides at the interface of a metal contact layer by the reaction of metal films when grown on silicon. We now illustrate how this thin film materials microstructure is analyzed by RBS. Figure 5.4.4a shows the 170◦ back-scattered spectrum of a 1000 Å thick film of Ni, deposited on a silicon substrate, probed with 4 He+ primary particles of energy, E0 = 2 MeV. The penetration depth of the 2 MeV 4 He+ beam in Ni is many microns. The particles back-scattered, with energy E 1 , from the front surface, satisfy the kinematic equation, (5.3.11), which gives E1 = K1 E0 = 1520 keV, with K1 = 0.76 for Ni, and K1 = 0.57 for Si. The 2 MeV particles traveling through

OUP CORRECTED – FINAL

322 Probes: Sources and Their Interactions with Matter (a)

(b) E0 Si

Ni

E Si Energy Si 2500Å

HNi

HSi

Ni

KSiE0

KNiE0

Ni 1000Å

Scattering yield

Scattering yield

E0

E0 Depth

Si

Si Energy

Ni2Si

ENi HNi

ESi HSi KSiE0 1500Å

KNiE0 1500Å

E0 Depth

Figure 5.4.4 Back-scattering yield for MeV 4 He+ ions from (a) 100 nm thick Ni film grown on a silicon substrate, and (b) a reacted Ni2 Si (silicide) layer. The energy scale is converted to depth as shown. Adapted from Feldman and Mayer (1986).

the nickel layer lose energy continuously along the path at the rate of 64 eV/Å (stopping power, S = 70.04 eV/(1015 atoms/cm2 ), and ideal atom density (NNi = 9.17 × 1022 atoms/cm3 ). Assuming the energy loss is linear with the thickness, the 2 MeV particles will lose 64 keV in energy while penetrating through the nickel layer. Just at the interface, the particles back-scattered by Ni atoms will have an energy KNi (E0 − 64) keV, and as they propagate outward they will continue to lose energy at the rate of 69 eV/Å (the difference from 64 eV/Å is due to the small energy dependence of the stopping power, S), and emerge with an energy of 1402 keV. Thus, the energy difference, ENi , of the particles emerging after backscattering from the Ni surface and the Ni/Si interface is, ENi ∼ 1520 − 1402 = 118 keV. On the other hand, if the Ni were to react with silicon and form a Ni2 Si compound (Fig. 5.4.4b), we can see that the energy width, ENi , is broader due to the presence of Si in the reacted layer. In addition, the back-scattering spectrum for Si shows a step corresponding to its presence in the Ni2 Si layer. Further, the product of the respective step heights, HNi and HSi , and energy widths, ENi and ESi , are proportional to the composition of the silicide layer, i.e. NNi HNi ENi σNi HNi ENi = = NSi HSi ESi σSi HSi ESi



ZSi ZNi

2 (5.4.12)

OUP CORRECTED – FINAL

Ion-Based Characterization Methods 323 Example 5.4.3: A thin film, 100 nm thick, of YBa2 Cu3 O7 is grown on a single crystal SrTiO3 substrate (adapted from Vickerman, 2000). (a) What would the RBS spectrum look like if 2.5 MeV 4 He2+ primary ions are used? (b) What would the individual components of the YBa2 Cu3 O7 spectrum look like after background subtraction? (c) Derive a simple expression for the compositional ratio of two elements, assuming that their energy widths are the same. Solution: (a) The expected RBS spectrum is plotted in the following figure.

Yield (100 counts/channel)

80

60 Cu O

40

Ti

Y

Sr

Ba

20

0 0

100 200 Energy (channel number)

300

Yield (100 counts/channel)

(b) The following figure shows what, after background subtraction, the individual component of the YBa2 Cu3 O7 RBS spectrum would look like. 60 O

40

Ba

Ti Sr

Cu Y

20

0 0

100 200 Energy (channel number)

300

OUP CORRECTED – FINAL

324 Probes: Sources and Their Interactions with Matter

(c) The stoichiometry of the film can be calculated by taking ratios of any two species, i and k, as in (5.4.12) to give   Ni Hi Ei σi Hi Ei Zk 2 = = Nk Hk Ek σk Hk Ek Zi

(5.4.13)

As a first approximation, we can assume that Ei = Ek , and obtain   Hi Zk 2 Ni = Nk Hk Zi

(5.4.14)

as an estimate of the stoichiometry.

5.4.1.4

Ion Channeling

In single crystal materials, a high-energy ion incident along certain crystallographic orientation encounters far fewer atoms, or scattering centers, when compared to other incident directions. This geometric effect is termed channeling and is exhibited by all incident probes—photons, electrons, neutrons, and ions. If the film is highly monocrystalline, and the collimated primary ion beam is incident along a low-index crystallographic direction, then it is steered in between the atomic columns (Fig. 5.4.5). This is called channeling, and in this case, the back-scattering yield is substantially reduced when compared to the case of

Figure 5.4.5 A schematic representation of the path of an ion when channeled in a diamond cubic lattice. Adapted from Bird and Williams (1989).

OUP CORRECTED – FINAL

Ion-Based Characterization Methods 325 Host Lattice (a) Substitutional

Substitutional impurity Face-centered interstitial YI

Body-centered interstitial

Random Surface

Directions 1 and 2

zi

zi Depth of Impurities

(c) Face-centered

(b) Body-centered

2

1 Random

Random

Surface

2

Surface

Direction 1

Direction 2

zi

1

Surface

zi

Figure 5.4.6 Determination of the position of foreign atoms in a lattice by the channeling and blocking of the primary high-energy ion beam incident along different directions. For the structure shown on the left and ions incident along the two directions, 1 and 2, the depth profile is shown for impurities on (a) substitutional sites, (b) BCC interstitial sites, and (c) FCC interstitial sites. Adapted from Bird and Williams (1989).

“random” incidence. However, if there are any deviations from this ideal crystal lattice, such as the presence of interstitial atoms, the back-scattered flux intensity and its angular distribution can vary. By appropriate triangulation, the position of foreign atoms in the lattice can be determined (Fig. 5.4.6).

5.4.2 Low-Energy Ion Scattering Spectroscopy (LEISS) A low-energy ion scattering spectroscopy experiment is similar to RBS but is carried out at low energies ( ∼keV). It also satisfies the kinematic relationship between energy and mass of the ions as described in (5.3.11) and (5.3.12). However, the scattered ions are detected using an electrostatic analyzer and therefore only positively charged ions are detected. For low-energy ions, only those particles back-scattered from the top surface monolayer have a high probability of surviving the scattering without their charge being neutralized. Thus, since the electrostatic analyzer requires a positive charge, LEISS is a highly surface-sensitive technique.

OUP CORRECTED – FINAL

326 Probes: Sources and Their Interactions with Matter Further, LEISS exhibits a sharp peak for each element present on the surface (Fig. 5.3.17); in comparison, a sharp edge in the spectrum for each element is observed in RBS. The sensitivity to neutralization is most pronounced for noble gas (He+ , Ne+ , + Ar ) ion scattering, and typically, the probability, P, for surviving as an ion after scattering is ∼5% for the first atomic monolayer, and at least one order of magnitude lower (∼0.5%) for deeper levels; hence, the surface sensitivity. However, the ion survival probability, P, is not well-known; typical values are ∼10% for 1 keV 4 He+ , ∼5% for 1 keV 20 Ne+ . Thus, quantitative analysis of LEISS data is not straightforward largely because of poor knowledge of the probability, P. Additionally, there is also substantial uncertainty in the absolute scattering crosssections for low-energy ions. Figure 5.4.7a shows a typical LEISS spectrum for an Al2 O3 surface covered with a monolayer of Rh. It is compared with an RBS spectrum (Fig. 5.4.7b) of the same film using 1 MeV 4 He+ ions. Notice the sensitivity of LEISS to the surface monolayer of Rh.

ISS

O Al

Rh

E0 = 500 eV

Fluence (1016He+/cm2) 36.9

O

Al

Rh ×15

28.7

19.5 14.6

Intensity (arb. units)

11.9 8.7

Yield (a.u.)

23.3

7.0 6.0 5.2 4.1 3.2

0

100 200 300 400 500 600 700 800 900 Backscattered ion energy (keV)

2.5 1.9 1.3 0.7 0.06

100

200

300 400 Energy (eV)

500

Figure 5.4.7 (a) LEISS and (b) RBS spectra from the same specimen of Al2 O3 with a surface monolayer of Rh. Adapted from Vickerman and Gilmore (2009).

OUP CORRECTED – FINAL

Ion-Based Characterization Methods 327 In summary, LEISS can be used to determine (1) the chemical species on the surface, (2) the relative concentration of different elements on a surface, and (3) where the surface atom (impurity, adsorbate) is located relative to the host lattice (how?). Example 5.4.4: Consider the ion scattering experiment shown in the following figure. 2 MeV, He2+

10°

Element, B Element, A

(a) Calculate the kinematical scattering factors for 79 Au197 , 14 Si28 .  ◦  (b) If necessary, make a very simple assumption that dE dz Si = 20eV / A ,  ◦  and dE dz Au = 40eV / A for both the incoming and outgoing ions. Now: (i) If element A is gold (Au) and the atoms of element B, dispersed on its surface are silicon (Si), plot its RBS spectrum. (ii) If the situation is reversed, i.e. element A is silicon (Si) and the atoms of element B, dispersed on its surface are gold (Au), plot the new RBS spectrum. (iii) Which of the two, (i) or (ii), would be better to resolve in an RBS experiment? Why? (iv) For (i) and (ii), if we perform a LEISS measurement with 2 keV incident ions, what would the corresponding spectra look like? (v) When would you prefer to use LEISS over RBS? Solution: (a) Kinematic factors, (5.3.11), for θ = 170◦ , with AAu = 197/4 = 49.25 and ASi = 28/4 = 7, are KAu = 0.923 and KSi = 0.565. (b) Thus, for E0 = 2 MeV, E1 (Au) = 1.85 MeV and E1 (Si) = 1.13 MeV.

OUP CORRECTED – FINAL

328 Probes: Sources and Their Interactions with Matter (i) Here Au is a thick film and its RBS spectrum would look like Figure 5.4.3, with an onset at 1.85 MeV. Si will have a sharp peak at 1.13 MeV, similar to Au in Figure 5.4.2. The overall RBS spectrum will be a sum of the two, with a small Si peak superimposed on the Au, as shown in the following figure. (ii) Si is the thick film and Au is on the surface. The onsets will be the same, but now Au and Si features are separately resolved. RBS

LEISS

Si Intensity (a.u.)

Au

Intensity (a.u.)

Au Si on Au film

Si

Au on Si film 1.85 1.13 Energy (MeV)

1.13 1.85 Energy (MeV)

(iii) Clearly, RBS will resolve Au atoms on Si film better than Si atoms on Au. (iv) LEISS mainly probes the surface atoms with intensity proportional to the concentration. The corresponding spectra for the two cases are shown in the figure on the right. (v) For Si on Au film, we prefer to use LEISS over RBS.

5.4.3

Secondary Ion Mass Spectrometry (SIMS)

When a solid is bombarded by ions, a number of interactive effects take place (Fig. 5.3.13), including changes in local chemistry (implantation) and structure (atomic displacements and sputtering). If the primary ions are incident normal to the surface of a specimen, the structural and chemical alteration of the solid extends to a depth equivalent to the projected range, Rp , of the ion (Fig. 5.3.14). After the passage of a certain period of time, corresponding to the sputtering of twice the projected range (depth = 2Rp ) of the specimen, an equilibrium of sorts between implantation and sputtering is established. From this point on, the composition of the materials removed by sputtering, as secondary ions, resembles closely the composition of the specimen, and can be analyzed to give useful information. However, the fraction of secondary ions in the sputtered material is quite small ( d. We shall derive a simple expression for the intensity at any point, P, on the screen, at a distance, z, as shown. The two spherical harmonic waves, emanating from S1 and S2 , travel with a path difference, lp (= S2 P—S1 P) and are then superimposed with a phase difference, , given by

= lp

2π λ

(6.4.1)

The intensity, (6.2.15), at P is then IP = 4a2 cos2 ( /2)

(6.4.2)

where a is the amplitude of the two spherical waves. Note that D >> d and D >> z. Under these conditions, the path difference lp ∼ d sin θ = d

z D

(6.4.3)

Thus, the intensity, (6.4.2), has a maximum value of 4a2 whenever is an integer multiple of 2π . From (6.4.1), these maxima will occur when lp ≈ mλ, where m is an integer, and bright fringes are observed when D d

(6.4.4)

dz = d sin θ D

(6.4.5)

z = mλ or mλ =

Similarly, we can show that the intensity is a minimum with a value of zero, whenever the phase difference = π , 3π , 5π . . . , and for these values corresponding to   D 1 λ z= m+ 2 d

5 Thomas Young (1773–1829) revived the Huygens wave theory. Also known for his work on elastic behavior of solids (Young modulus).

(6.4.6)

we observe dark fringes. Figure 6.4.1c,d shows such an interference pattern for a double-slit experiment. One of the principal requirements for the interference of two beams to produce a stable pattern is that they must have (very nearly) the same frequency. If the frequency difference is large, it would result in a time varying phase difference, which would then cause the resultant intensity to average to zero.

OUP CORRECTED – FINAL

Reflection and Refraction 359 (c) D P S1 S

d

θ

θ

z

S2 A

(d) 5π 4π 3π 2π π 0 π –2π –3π –4π –5π

I

4a2

(b)

2a2

(a)

Figure 6.4.1 Young double-slit experiment: (a) A plane wave passes through a pin hole, S, and then after a certain distance passes though (b) two pin holes and the resultant projected on a screen. The path difference, d sin θ , between the waves from S1 and S2 , leads to constructive and destructive interference as a function of position, P, on the screen. (c) The observed fringe pattern and (d) the intensity, I, variation as a function of the phase difference, Δ, for the first few fringes are also shown.

Example 6.4.1: In a Young double-slit experiment, the two slits separated by 1.0 mm are illuminated with green light (λ = 550 nm), and the interference pattern is observed on a screen placed at a distance of 1.0 m from the slits. What is the separation between two successive bright or dark fringes? Solution: The separation between the bright fringes (dark fringes will be the same) are given by (6.4.5). Here, d = 1.0 mm, D = 1.0 m, λ = 550 nm, and fringes will be observed for z = mλ D/d = m 550 × 10−9 /10−3 = m 5.5 × 10−4 m = 0.55m mm. Thus, for m = 1, the separation between successive fringes will be 0.55 mm, and may require magnifying glasses to be observed by the eye.

6.5 Reflection and Refraction Huygens’ principle can be readily applied to well-known laws of reflection and refraction of light (Fig. 6.5.1) by simply considering point sources on the interface of mediums 1 and 2, and then deriving the resultant wave in the new medium (refraction) or direction (reflection). We first consider reflection (Fig. 6.5.1).

OUP CORRECTED – FINAL

360

Optics, Optical Methods, and Microscopy B A θ1 θ 1

OB

P

Medium 1 R

O Figure 6.5.1 Reflection of light from a surface separating Medium 1 and Medium 2.

θ2 S Medium 2

(a) B (b)

A v1t θ1

1 2

Surface

O

R

Medium 1 Medium 2

v2t

Medium 2

3

Medium 1

θ2 θ2

3

2

1

A Figure 6.5.2 (a) Refraction of light at the interface between two media, and (b) total internal reflection (of ray 3).

At time, t = 0, the ray AO reaches the surface of medium 2, but the wave at B θ1 will reach R later at a time tB = OvB1R = ORvsin . The spherical wave leaving O, 1 OR sin θ1

will reach P, at a time tP = OP v1 = v1 for the two waves to be in phase, then

. If the two times are equal, i.e. tB = tP ,

sin θ1 = sin θ1

(6.5.1)

or the angle of incidence is equal to the angle of reflection, i.e. θ1 = θ1 . Generally, a fraction of the light will penetrate medium 2, from medium 1, and be refracted (Fig. 6.5.2a); the rest (not shown) is reflected at the interface. Here, the velocity of light in the two media are, v1 and v2 , respectively. In medium 1, ray

OUP CORRECTED – FINAL

Diffraction 361

Dense Flint Glass

Index of Refraction, n

1.7

Light Flint Glass 1.6 Crystal Quartz 1.5

Acrylic Figure 6.5.3 The change in index of refraction, n, as a function of the wavelength of light for various materials. The visible region is highlighted.

Fused Quartz 1.4

0

200

400

600

800

1000

Wavelength, λ(nm)

Adapted from Hecht (2001).

A will reach point O first, and then after a time, t, ray B will reach R. In this time, ray B travels a distance v1 t = OR sin θ1 in medium 1, and ray A travels a distance v2 t = OR sin θ2 in medium 2. Thus, we get the Snell6 law of refraction: v1 n2 sin θ1 = = sin θ2 v2 n1

(6.5.2)

where we define the refractive index, n, of the medium as the ratio between the speed of light in vacuum to the speed of light in the medium. Note that the velocity of light in a specific medium also depends on its wavelength (Fig. 6.5.3) for various optical materials, including the commonly used fused quartz. Further, if light is traveling from a medium of higher refractive index to a medium of lower refractive index, i.e. n1 > n2 , it is bent away from the surface normal towards the plane of the surface. Thus, at some incident angle, θ1 = θc , we reach a point where n1 sin θc = n2 sin (π/2) = n2 , or θc = sin−1 (n2 /n1 ) and all light is reflected (no refraction) for θ1 > θc (Fig. 6.5.2b). This is known as total internal reflection and is used in many applications. In particular, it is used in fiber optic cables to transmit light signals without any loss in signal intensity.

6.6 Diffraction When light passes through a narrow slit or pin hole it is diffracted (Fig. 6.3.1). In fact, this is a general characteristic of wave phenomena and occurs whenever a portion of a wave front is obstructed in some way. Figure 6.6.1 shows a common example of this phenomenon that deviates significantly from geometric optics.

6 Attributed to the Dutch mathematician, W. Snellius (1580–1626), but historical research shows that this law already was formulated by the Persian mathematician and optician Ibn Salah in 984 CE!

OUP CORRECTED – FINAL

362

Optics, Optical Methods, and Microscopy

Figure 6.6.1 The shadow of a hand holding a coin illustrating the effect of diffraction.

(a)

(b)

Aperture

Aperture Figure 6.6.2 Diffraction by an aperture or slit, illustrating the conditions for (a) Fraunhofer and (b) Fresnel diffraction. Here, S is the source and P is the point of detection.

S To S To P

P

Adapted from Fowles (1989).

7

Joseph Fraunhofer (1787–1826).

8

Augustin J. Fresnel (1788–1827).

In the detailed treatment of such diffraction, it is common practice to distinguish between two general cases known as Fraunhofer7 and Fresnel8 diffraction (Figure 6.6.2). Fraunhofer diffraction is important to understand the behavior of gratings, introduced in Figure 3.5.1 in the context of infrared spectroscopy, and is discussed further in §6.6.3. It is expected when the incident and diffracted light are essentially plane waves; in practice, this is the case when the distances from the source to the diffraction aperture and from the aperture to the screen are sufficiently far apart to neglect the curvatures of the incident and diffracted waves (Fig. 6.6.2a). On the other hand, if either the source or the detector is close enough to the diffraction aperture, such that the curvature of the wave front is significant, we have Fresnel diffraction (Fig. 6.6.2b). A practical application of Fresnel diffraction is the design of zone plates used routinely to focus high energy X-rays (see §6.6.7). We shall now discuss specific examples of Fraunhofer and Fresnel diffraction by various types of apertures.

OUP CORRECTED – FINAL

Diffraction 363

6.6.1

Fraunhofer Diffraction from a Single Slit

Figure 6.6.3a shows the experimental arrangement for Fraunhofer diffraction from a single slit of width, w. Figure 6.6.3b shows the geometrical construction to obtain the intensity distribution on a screen at a distance, D (where D >> w), illuminated by a plane wave. For analysis, the slit is divided into infinitesimally small elements of width, ds, at a distance, s, from the origin, O, each of which is a source of secondary wavelets. For the wavelet emitted by the element, ds, at the origin, its amplitude at point. P, on the screen will be directly proportional to the element width, ds, and inversely proportional to the distance, x. Thus, the displacement, dz0 , it produces at the point, P is dz0 =

a ds sin (ωt − kx) x

(6.6.1)

For the element, ds, at the position +s, the difference in path-length, lp , is given by lp = s sin θ . The corresponding displacement for this wavelet is dzs =

  ads ads sin ωt − k x + lp = sin [ωt − k (x + s sin θ)] x x

(6.6.2)

and the total displacement is obtained by integrating (6.6.2) to obtain the contributions of all the wavelets emanating from all the elements of the slit, i.e. z=

+w/2 −w/2

dzs =

+w/2 −w/2

a sin [ωt − k (x + s sin θ )] ds x

(6.6.3)

Following integration and rearrangement of terms we get

z=

2a sin [ωt − kx] x

= A0





w/2

cos (ks sin θ ) ds =

0

aw sin x

sin β sin [ωt − kx] = A sin [ωt − kx] β

kw 2 sin θ kw 2 sin θ

 sin [ωt − kx] = (6.6.4)

Thus, the important physical result is that the resultant displacement will also be a simple harmonic motion, determined by θ , but which will vary with position. The amplitude, A = A0 sinβ β , where A0 = aw/x and β = kw 2 sin θ = πw sin θ . Note that at β = 0, A = A , we observe a maximum value, and 0 λ at β = ±π, ±2π , . . . , A = 0, we observe minimum values. The intensity, I, on the screen is I = A2 =

A20 sin2 β β2

(6.6.5a)

OUP CORRECTED – FINAL

Optics, Optical Methods, and Microscopy (a) L1

S

P.

L2

S

Source Source Slit

Diffraction Slit

(b)

Screen

3π

P

2π

x –π π

ds

P0

θ lp

2π 3π

Amplitude

s

0

θ

wO

Intensity

364

D ( >> w) β Figure 6.6.3 (a) Experimental arrangement to obtain the Fraunhofer diffraction pattern from a single diffracting slit. (b) Geometrical construction to determine the intensity distribution of the wave whose amplitude and intensity variations are also shown.

Both the amplitude and intensity variation, as a function of β, in units of radians, are also plotted in Figure 6.6.3b. Now, if the plane wave is incident at some angle, φ, then β is given in its more general form, i.e. β = πw λ (sin θ + sin φ), and includes both θ and φ. Further, if the slit has a finite size, i.e. width, w, and height, h, then its amplitude sin β sin γ πw πh Aw×h = awh x β γ where β = λ sin θ and γ = λ sin, where θ and  define the angles of the diffracted ray in two orthogonal planes. Its intensity, I, is given by  I = A2w×h = I0

sin β β

2 

sin γ γ

2 (6.6.5b)

OUP CORRECTED – FINAL

Diffraction 365 I/I0

(a)

(b) h

Figure 6.6.4 (a) Fraunhofer diffraction from a square slit (h = w), showing the intensity variation along the two orthogonal directions. (b) An actual diffraction pattern of a rectangular slit with h = 2w; in the latter, the spacing in the horizontal and vertical directions are different and determined by the dimensions of the slit.

w γ 2π π –2π

π

–π

2π

β

–π –2π

(a) Adapted from Hecht (2001). (b) Adapted from Jenkins and White (1981).

I/I0

ρ

Airy disk Figure 6.6.5 Fraunhofer diffraction from a circular aperture. The variable, ρ = kR sin θ , where R is the radius of the aperture, and k is the magnitude of the wave vector of the incident wave.  2 . Figure 6.6.4b shows the Fraunhofer diffraction pattern of where I0 = awh x such a rectangular aperture (h = 2w). If instead of a rectangular slit, a circular aperture is used, a similar integration has to be carried out over the two-dimensional surface. The problem was originally solved by Airy9 in 1835, with the solution obtained in terms of Bessel functions. Details can be found in any advanced book on optics (Fowles, 1989; Hecht, 2001), but the diffraction pattern obtained with such a circular aperture (Fig. 6.6.5) is also circularly symmetric, with a bright central disc—

9 George B. Airy (1801–92) was an English mathematician and astronomer.

OUP CORRECTED – FINAL

366

Optics, Optical Methods, and Microscopy known as the Airy disc—surrounded by concentric bands of diminishing intensity. Alternatively, using a simple symmetry argument, the diffraction pattern can be generated by rotating the diffraction pattern obtained from a single slit (Fig. 6.6.3) about its axis. The angular radii of the finite disc are given by sin θ = 1.22 λ/D, where D = 2R, is the diameter of the aperture.

6.6.2 Fraunhofer Diffraction from Double and Multiple Slits Consider Figure 6.6.6 to observe the Fraunhofer diffraction pattern from a double slit, where the two slits of width, w, are separated by a distance, d. Following the derivation for the single slit, we can show that the displacement on the screen is given by

z=2

aw sin β cos γ sin [ωt − kx] x β

πd where β = πw λ sin θ , γ = λ sin θ , and A0 =

I = 4A20

aw x .

(6.6.6)

The intensity is given by

sin2 β cos2 γ β2

(6.6.7)

which shows positions of maxima, with values of Imax = 4A20 determined by cos2 γ at γ = 0, π , 2π ....., or equivalently when d sin θ = nλ, where n is an integer. Note that the factor, (sin β/β)2 , derived earlier (6.6.5a) for the single-slit appears as an γ

β 2π

S Source Figure 6.6.6 Fraunhofer diffraction from a double-slit aperture, including the intensity distribution. Note that the effect of the slit width, β, modulates the effect of their separation, γ .

b c b

d

+π +2π +π 0 0 –π –2π

θ

–π Double Slit –2π I/I0

OUP CORRECTED – FINAL

Diffraction 367 (a)

(b) I/I0 5 slits

1 slit

I/I0 0

5 slits

π

2π

γ

3π 20 slits

2 slits

0

6 slits

π

2π

γ

3π

(c)

3 slits

0th order

1st order

2nd order

3rd order

4th order

5th order

θ

Figure 6.6.7 (a) Fraunhofer diffraction patterns from gratings with different number of slits. The sharpness of the principal maxima increases with the number of slits. (b) The intensity distribution plotted for 5 and 20 slits, (c) The separation of two different wavelengths as a function of diffraction order, m = 0, 1, 2, 3, 4, . . . . Adapted from Jenkins and White (1981).

envelope of the diffraction pattern. The resultant intensity, I, will be zero if either β = π, 2π, 3π, 4π, . . . or γ = π/2, 3π/2, 5π/2 . . . , as shown in Figure 6.6.6. Example 6.6.1: What happens to the diffraction pattern for a double slit if the width, w, is equal to half the spacing, d, i.e. w = d/2? Solution: The angle at which the maxima of the diffraction occurs will be at sin–1 (nλ/d). However, the angle at which the minima of the envelope occurs will be at sin−1 (nλ/w) = sin−1 (2nλ/d). Thus, every other maximum (n = 2, 4, 6, . . . ) in the diffraction pattern will be missing.

As the number of slits, N, is increased from two (Fig. 6.6.7), we can see that the sharpness of the principal maxima increases rapidly with the number of slits, and for N = 20, the lines become very narrow indeed. Further, secondary maxima appear as shown for N = 3 and N = 5 slits. Now we can generalize (without proof ) our analysis to N slits, each of width, w, and separated by distance, d, to give an intensity

OUP CORRECTED – FINAL

368

Optics, Optical Methods, and Microscopy IN = A2 = A20

sin2 β sin2 Nγ β 2 sin2 γ

(6.6.8)

πd aw Recall that β = πw λ sin θ , γ = λ sin θ , and A0 = x . Further, when N = 2, we retrieve the result for the double slit, (6.6.7), and in the general case, maximum values, INmax = N 2 A20 , are expected for γ = 0, π , 2π ..... The new term, (sin2 Nγ )/(sin2 γ ), represents the interference effect for N slits. Note that

N cos Nγ sin Nγ lim lim = = ±N γ → mπ γ → mπ sin γ cos γ

(6.6.9)

Thus, the maxima at γ = 0, π , 2π ..... correspond exactly in position to the maxima observed for a double slit as these conditions for γ also lead to the condition, d sin θ = mλ, where m is an integer. However, INmax = N 2 A20 , and the intensity is proportional to the square of the number of slits. Further, for a given order, m, the angular separation (Fig. 6.6.8) between two spectral lines differing in wavelength, λ, is given by

θ =

m λ d cos θ

(6.6.10)

The secondary maximum also follows a pattern that is related to the number of slits. As the number of slits increases, the amplitude of the secondary maximum decreases (Fig. 6.6.7). Fraunhofer diffraction from a grating with multiple slits provides the basis for a physical understanding of the diffraction of X-rays and electrons by crystalline materials, to be discussed in detail in §7 and §8, respectively. A diffraction grating is in essence a one-dimensional crystal that has three distinct variables: slit spacing, slit width, and number of slits. The slit spacing, d, is equivalent to the lattice spacing, and determines the directions or overall geometry of the diffracted beams. The effect (β) of slit width, w, modulates the effect (γ ) of their spacing. Analogously for X-rays, the atoms occupying the motif determine the observed intensities. Here, we assume that the diffraction is not dynamical and the diffracted beams do not interact with one another. Finally, the total number of slits determines the number and intensities of subsidiary diffraction peaks (satellites) on either side of the main diffraction peak. In fact, the number (intensities) of satellites increase (decrease) with the number of slits. For bulk crystals, the number of lattice planes involved in X-ray diffraction is very large, and as a result the satellite peaks have no observable intensity. However, thin films and multilayer materials with finite extent and a limited number (of the order of 100 or 1000) of lattice planes are an exception, and show well-defined satellite peaks that can be helpful in their analysis. Figure 7.9.12 shows a practical example.

OUP CORRECTED – FINAL

Diffraction 369

6.6.3

Resolving Power of a Diffraction Grating

The angular width of a principal fringe (Fig. 6.6.6), defined as the separation between a peak and its adjacent minimum, is obtained by setting the change in Nγ equal to π, i.e. when γ = π/N = πd λ cos θ θ , or when

θ =

λ Nd cos θ

(6.6.11)

Thus θ is inversely related to N, and Figure 6.6.7c shows how, when N is very large, we observe sharp fringes in the diffraction pattern corresponding to different orders, m = 0, ±1, ±2, ..... From (6.6.10) and (6.6.11), we can show that the resolving power (RP) of the grating, defined as λ/ λ, is given by R.P. = λ/ λ = Nm

(6.6.12)

This is illustrated in Figure 6.6.8 and in Figure 6.6.7c. Note that the resolving power is not dependent on the size, w, of the individual grating (or ruling) or their spacing, d; it only depends on the total number, N, of gratings (rulings) and the order, m, of the diffraction. For various optical spectroscopies, discussed in §3, reflection gratings are typically used. They contain 600 lines/mm over a total width of 10 cm, giving a total of 60,000 lines, and a resolving power of 60,000m, where m is the order of the diffraction. By suitably shaping the grooves, e.g. giving them a saw-tooth profile, the majority of diffracted light can be made to appear in one order, thus increasing the efficiency of the grating. For this to be successful, the spacing between the lines has to be very uniform, and moreover, its magnitude must be much smaller than the wavelength.

Incident Light (plane wave)

(a) Nm

λ

θ λ

θ

Diffracted Light

mλ

(b)

Nd d θ

θ

B θ

Grating with multiple (N ) slits

Figure 6.6.8 (a) A transmission diffraction grating arrangement to resolve spectral lines. (b) The angular separation of two spectral lines,

θ , just resolved by the diffraction grating. Adapted from Jenkins and White (1981).

OUP CORRECTED – FINAL

370

Optics, Optical Methods, and Microscopy Example 6.6.2: The optical path through a transmission diffraction grating is shown in the following figure.

A B n

θi

a

θn C

D

θm

Show that the optical path length difference is independent of the refractive index of the grating. Solution: From (6.5.2), sin θi = n sin θn , where n is the refractive index of the grating. The optical path difference = CD − AB = mλ = a (n sin θn − sin θm ) = a (sin θi − sin θm ) , and does not depend on the refractive index.

6.6.4

Fresnel Diffraction

Fresnel diffraction can be easily understood by referring to Figure 6.6.9, where wave (1) starts at the bottom edge of the slit, and wave (2) starts at the middle of the slit. Note that the width, d, of the slit is much smaller than the distance, D, of the slit from the screen, i.e. d D, and no lenses are required. The path difference, x, between rays (1) and (2), at the point, P, located at an angle, θ, is

x = (d/2) sin θ , with an intensity minimum when x = λ/2. If we choose θ in such a way that (d/2) sin θ = λ/2, then every ray from the bottom half of the slit will be cancelled by another ray from the top half of the slit. Thus, the resultant intensity at this position, defined by θ , on the screen will remain zero and we get the first minimum. Similarly, even for a single slit, higher order minima are observed when d sin θ = nλ, where n is an integer.

Slit B Figure 6.6.9 Conditions for Fresnel diffraction. Note that, in reality, d D.

P

(2) θ (1)

Source d

A

D

OUP CORRECTED – FINAL

Diffraction 371

6.6.5

Fresnel Half-Period Zones

Consider the divergent spherical wave front, outlined by BCDE, traveling to the right (Fig. 6.6.10a). Following Huygens principle, every point on this spherical surface can be thought of as a source of secondary wavelets. To determine the effect of these wavelets at the point, P, of interest at a distance, OP = b, from the center, we divide the wave front, BCDE, into circles, at distances, s1 , s2 , s3 , ...., along the arc, as shown, such that each of these circles are one half of a wavelength further from P, i.e. the circles are at distances b + λ/2, b + 2λ/2, b + 3λ/2, . . . from P. The areas, Sm , of the circles or zones, are all effectively equal, as can be seen in the simple construction shown in Figure 6.6.10b. The path difference, lp = HQP − HOP, must be a multiple of λ/2 at the borders of the zone. From the sagittal formula,10 for any zone with radius, s, we get lp =

s2 s2 a+b + = s2 2a 2b 2ab

(6.6.13)

giving the radii, sm , of the Fresnel zones as s2 λ a+b == m m = s2m 2 2ab 2



1 1 + a b

 (6.6.14)

with the area of any one zone being the difference in the areas of two adjacent circles.   a λ 2ab Sm = π s2m − s2m−1 = π = π bλ 2 (a + b) a+b

(6.6.15)

In practice, the area of each zone is independent of m, and to first order a constant. Thus, by definition, each zone will send out successive wavelets that differ by a phase of π at P. Their amplitude, Am , given by

B

(a) E

s3 s2 s1

Am = k

Sm (1 + cos θ ) dm

b+3λ/2 b+2λ/2 b+λ/2

(b)

(6.6.16)

Q lp a

H

O

b

P

R b

s

P

O a

b

C D

10

The sagittal formula states that the distance, s, from the center of a circular arc of radius, l2 r, to its base of length, l, is given by r = 2s + 2s .

Figure 6.6.10 (a) A spherical wave front divided into half-period zones. (b) The path difference, , at the point, P, from a distance, s, from the pole of the spherical wave. Adapted from Jenkins and White (1981).

OUP CORRECTED – FINAL

372

Optics, Optical Methods, and Microscopy where k is a constant, dm = b + lp , is the distance of each zone to P, and θ is the angle at which the light leaves the zone, will add in a series with alternative signs to give a resultant amplitude, A, as A = A1 − A2 + A3 − A4 + A5 + ....(−1)m+1 Am

(6.6.17)

since successive zones differ by a phase of π. In (6.6.16), (1 + cosθ) is an “obliquity” factor that causes successive terms in (6.6.17), to decrease very slowly. We can rearrange (6.6.17) as A1 A= + 2



A3 A1 − A2 + 2 2



 +

A5 A3 − A4 + 2 2

 + ..... ±

Am 2

(6.6.18)

and because the amplitude for neighboring zones are nearly equal, we get for odd m, the total amplitude at P, as A=

Am A1 + 2 2

(6.6.19a)

A=

A1 Am − 2 2

(6.6.19b)

However, if m is even, we get

In summary, the resultant amplitude at P, is either half the sum (odd m) or half the difference (even m) of the contributions of only the first and last zones. Further, if m is very large, the entire spherical wave front when divided into zones results in θ → 180◦ for the last zone. Thus, the obliquity factor (1 + cosθ ) in (6.6.16) becomes negligible and Am → 0, A → A1 /2, i.e. the amplitude of the whole wave is just half of the first zone acting alone.

6.6.6

Diffraction by a Circular Aperture or Disc

For a spherical wave encountering a circular aperture (Fig. 6.6.11a) with radius, r = s1 , where s1 was defined earlier (Fig. 6.6.10a), we get from (6.6.19a) for the total amplitude A=

A1 A1 + = A1 2 2

(6.6.20a)

A1 A2 − ≈0 2 2

(6.6.20b)

However, if r = s2 , we get A=

OUP CORRECTED – FINAL

Diffraction 373 (a)

(b)

R r O

P

Figure 6.6.11 (a) Light passing through an open slit showing the geometric arrangement. (b) The Fresnel diffraction observed from a circular opening. Adapted from Jenkins and White (1981).

Thus, the intensity will oscillate with the size of the hole. Also, the path difference will vary along the axis of the aperture and the intensity will vary as a function of distance, b. From the previous, as a corollary, if the spherical wave is diffracted by a circular disc, the resulting amplitude will again be a summation, as discussed for the circular aperture, but since the first few zones are blocked, they are omitted from the summation. This leads to a bright spot in the center of the shadow image and the circular disc acts as crude lens.

6.6.7 Zone Plates and Their Applications in X-Ray Microscopy Consider the special case of a half-period zone (§6.6.5), called a zone plate, where the transmitted light from every other half-period zone is blocked. As a result, either all the positive or all the negative terms in the summation (6.6.17) are removed. Either way, the amplitude, A, at P is increased many times. From √ (6.6.14), the radii of the zones, sm , satisfies the condition sm ∼ m for any given a, b, and λ. Thus, in practice, by drawing circles with successive radii proportional √ to m, where m is an integer, blackening alternative (either odd or even) zones, and allowing the light to pass through the rest, a zone plate can be constructed (Fig. 6.6.12a). Such a zone plate will produce an intense spot on its axis at a distance corresponding to the radii of the zones and the wavelength of the light used. The spot is so intense that the zone plate effectively functions as a lens. We illustrate this concept with some numbers. Assume that the first five odd zones are exposed. The effective amplitude, (6.6.17), is A = A1 + A3 + A5 + A7 + A9 = 5A1 . Without the zone plate, for large m, the whole wave gives an amplitude of A1 /2; thus, we can generalize and

OUP CORRECTED – FINAL

374

Optics, Optical Methods, and Microscopy (b) r = 25 nm, D = 63 μm, N = 618 zones (a)

Figure 6.6.12 (a) Zone plates. (b) Details of a zone plate fabricated for focusing an X-ray at a synchrotron beam line. Notice the period (Δr ∼ 25 nm) and the number, m = 618, of zones. (c) A schematic of a transmission X-ray microscope with zone plates as condenser and objective for imaging.

(c) Source

Condenser Plane Mirror

Objective

Condenser Zone Plate Pinhole Micro Zone Plate

ALS Bending Magnet

Specimen

Soft X-ray CCD

Figures courtesy of the Advanced Light Source, Berkeley.

say that by exposing m zones, only odd or even, we get an amplitude at P that is 2m times larger than that without the zone plate. Moreover, the intensity at P is 4m2 times greater. In addition, from (6.6.14), the source and image distances follow the standard lens formula (6.8.3) as 

1 1 + a b

 =

1 mλ = 2 f sm

(6.6.21)

with the focal length, f, given by the value of b when a → ∞, i.e. f =

s2 s2m = 1 mλ λ

(6.6.22)

In practice, such zone plates are lithographically fabricated of alternating rings of light (e.g. silicon) and heavy (e.g. tungsten) elements, and find great utility in focusing high-intensity X-rays such as in synchrotron beam lines. Figure 6.6.12b shows a typical zone plate fabricated by careful e-beam lithography and containing 618 zones. It is commonly used as a lens for a transmission X-ray microscope (Fig. 6.6.12c) in dedicated beam lines at synchrotron sources. Further details of zone plates and their use in imaging can be found in Atwood (2000).

OUP CORRECTED – FINAL

Visually Observable: Characteristics of the Human Eye 375 Example 6.6.3: A zone plate has a focal length of 0.5 m for a wavelength of 500 nm. What is the radius of the first and sixteenth circles of the zone plate? Solution: The radii, sm , of the Fresnel zones can be calculated from (6.6.22), i.e. sm = ( f m λ)1/2 .  1/2 Thus, the radius of the first circle, s1 = ( f λ)1/2 = 0.5 × 500 × 10−9 = 5 × 10−4 m. And the radius of the sixteenth circle, s16 = 4 s1 = 2 × 10−3 m.

6.7 Visually Observable: Characteristics of the Human Eye We now move from diffraction to discuss optical imaging and microscopy. For any visual observation, we have to first understand the characteristics of the human data collection and image recording system. So, we begin with a brief summary of the characteristics of the human eye to answer the question: “what is visually observable?” The human eye (Fig. 6.7.1) is a lens system that projects a positive image on a light sensitive surface (retina). The amount of light entering the eye is determined by a diaphragm (iris), which controls the light entering the eye through a variable hole (pupil). Two types of photo detectors are used to create black and white Iris

Retina Choroid

Cornea

Anterior Chamber Optic Nerve

Posterior Chamber Lens

Ciliary Muscle Figure 6.7.1 The human eye showing the principal optical components. Vitreous body

Adapted from Anatomica, Global Book Publishing (2005).

OUP CORRECTED – FINAL

376

Optics, Optical Methods, and Microscopy (rods) and color (cones) images. We have already remarked that the human eye is sensitive to wavelengths (visible spectrum) ranging from 380 nm to 700 nm, i.e. colors from violet to dark red (Fig. 1.3.2). However, the peak sensitivity of the human eye corresponds to green light with wavelength, λ = 550 nm. In practice, this is also one of the characteristic emission peaks (λ = 546 nm) of a mercury arc lamp (Fig. 5.2.1). Further, to take advantage of this peak sensitivity of the human eye, optical microscopes are focused using a green filter, and viewing screens for electron microscopes, both transmission and scanning, were traditionally coated with a green phosphor. Another important characteristic of the human eye is its integration time, which is the time required for the retina to collect sufficient photons to form an image. Typically, this is ∼0.1 s. In addition, the ambient light environment also affects the performance of the human eye. In total darkness the eye sees occasional flashes of light arising from random noise, but once it is adapted to the darkness, referred to as dark adapted, it achieves maximum sensitivity at low light levels and registers a good 50% of the incident photons! A reasonably good image can then be formed with ∼100 photons collected by each picture element, or pixel. Thus, the human eye is a remarkable “instrument” and performs as well as the best available night vision system, with one exception being that the latter can integrate signals for times longer than 0.1 s. Last, but most importantly, we are interested in determining the resolution of the eye, or its ability to separate two features that subtend the smallest angle at the eye, for a given pupil diameter, or aperture, at a specific distance of the objects from the eye. We address resolution in the next section.

6.8 Optical Microscopy 6.8.1

Resolution: Rayleigh and Abbe Criteria

Following the previous discussion of the Airy disc (Fig. 6.6.5), Abbe11 showed that a point source subtending an angle, 2θ, at the lens will have an apparent diameter, ∂, on the screen given by ∂=

1.2λ n sin θ

(6.8.1)

11

Ernst Abbe (1840–1905) was a German physicist. 12 John William Strutt, Lord Rayleigh (1842–1919) made fundamental discoveries in acoustics and optics that are basic to the theory of wave propagation. He received the Nobel prize in Physics (1904) and was cited for “investigations of the densities of the most important gases and for his discovery of argon in connection with these studies.”

where λ is the wavelength of the light and n is the refractive index of the medium (Fig. 6.8.1a). Thus, an object with a diameter, D, will have a size D + ∂ on the screen. Rayleigh12 then introduced a simple criterion for the resolution of the lens, by defining it as the separation of two point sources where the maximum intensity from one source coincides with the minimum intensity of the other (Fig. 6.8.1b). Hence, its resolution, δ = ∂/2 and is given by δ=

0.6λ n sin θ

(6.8.2)

OUP CORRECTED – FINAL

Optical Microscopy 377 (a)

(b)

δ=

0.6 λ n sinθ

Noise SNR~10%

In addition, note that the signal intensity has to be well above background noise; typically signal to noise ratio, SNR ∼10, is required for detection. Example 6.8.1: The human eye has a normal aperture (pupil) of ∼1.2 mm and it is physically difficult to focus it closer than ∼15 cm (called the near limit). What is the resolution of the eye when green light (λ = 550 nm) is used? Solution: For the eye, sin θ = 0.6/150 0.004, and for green light, λ = 550 nm, in air (n = 1), we get a resolution δeye = 0.6×550 1×0.004 82500 nm 82.5μm ∼ 0.1 mm.

Example 6.8.2: What is the optimal magnification for an optical microscope (Fig. 6.8.5)? Solution: The best optical systems are designed such that n sin θ = 1. Thus, its resolution is δmicroscope ≈ 0.6λ 330nm. For the eye to be able to optimally see a magnified image, it should satisfy the condition δeye = δmicroscope M. The optimal magnification for an optical microscope is Mopt δeye /δmicroscope 0.1 mm/330 nm ∼ 300. In other words, an optical microscope with magnification, M ∼ 300 – 400 will reveal all the information the human eye can resolve. Higher magnifications, M, do not bring any significant advantage, and using M > 500 for an optical microscope is often a waste of resources.

Figure 6.8.1 (a) The image of a point object at infinity is given by the Abbe criterion. The width of the first peak is a function of the angular aperture, θ, of the lens and the wavelength of light, λ. (b) The Rayleigh criterion for resolution is given in terms of the image of two point-objects at infinity. The resolution, δ, is defined such that the angular separation of the two images gives the maximum of one peak at the position of the minimum of the other.

OUP CORRECTED – FINAL

378

Optics, Optical Methods, and Microscopy

6.8.2

Geometric Optics and Aberrations

A brief discussion of elementary geometric optics is required to understand the operating principles of an optical microscope. As seen (Fig. 6.5.2), when light passes from a medium of low refractive index (e.g. air) to a medium of higher refractive index (e.g. glass) it is deflected, and the angles of transmission, for a given incidence angle, are determined by their respective refractive indices. Thus, a parallel beam of light, incident on a convex glass lens, is deflected such that the angle of deflection depends on the distance from the optic axis, and results in the light being focused at the focal point of the lens (Fig. 6.8.2a). The distance, f, of the focal point along the optic axis is proportional to the wavelength of light used. Further, simple geometric optics (Fig. 6.8.2b) determines the imaging characteristics and the magnification of the image. The distance of the object and image are simply related as −

1 1 1 + = u v f

(6.8.3)

and the magnification, M, is defined as the ratio v/|u|. 6.8.2.1

Lens Defects: Aberrations, Distortions, and Astigmatism

The lenses used in optical (and also other electromagnetic radiation) imaging do not give perfect images and are characterized by two important aberrations, chromatic and spherical (Fig. 6.8.3). Consider light emitted from the position P (Fig. 6.8.3a) with two different wavelengths, λ and λ + λ. As mentioned, the greater the wavelength, the greater is the focal length, and hence, these two wavelengths are brought to focus at two different points, Q and Q , on the optic axis. As a result, visible “white” light, consisting of a range of colors or wavelengths, are not brought to focus at a fixed point, but will be dispersed. Thus, point P, now imaged as a disc of radius R/M, where M is the magnification, defines the chromatic aberration coefficient Cc ∝ ( λ/λ) / R

(6.8.4)

Note that in the case of electrons (see §9.2.2), λ/λ is replaced by v/v, where v is the velocity, and (6.8.4) remains unchanged. (a) Figure 6.8.2 (a) A parallel beam of light incident on a convex lens converges on the focal point, at a fixed distance, f, along the optic axis. (b) Simple ray tracing showing image formation in a convex lens.

(b) f –u Focal Point Focal length, f

–f

v

OUP CORRECTED – FINAL

Optical Microscopy 379 (a) Gaussian Image Plane λ,λ+λ α

Q’

R A

P

λ+λ

λ

Q

(b) Gaussian Image Plane B α S

A

Q’

Q R

Paraxial Beam Non-paraxial Beam

(c)

Object

Pin Cushion Distortion

Barrel Distortion

Spiral Distortion

Figure 6.8.3 The two principal defects or aberrations commonly observed in both optical and electromagnetic lenses. (a) Chromatic aberration: rays leaving the same point, P, but with different wavelengths, λ and λ + λ (or velocities, v and v + v) are brought to focus at different points, Q and Q . As a result, the image of the point appears as a disc of radius, R, on the Gaussian image plane. (b) Spherical aberration: Paraxial and non-paraxial rays are brought to focus at different points along the optic axis. (c) Other commonly observed distortions in both optical and electromagnetic lenses. Adapted from Flewitt and Wild (2003).

The focal length of the lens varies across the lens (Fig. 6.8.3b). For paraxial13 illumination, all rays originating from P form images on the image plane at Q. However, if the rays are not paraxial, the rays are bent further at the extremities of the lens, and displace the image by a distance, R/M, to Q . Again, the point source, P, has an apparent radius R/M, which determines the spherical aberration coefficient, Cs , of the lens. Further, the resolution, δ, of the lens is now related to the spherical aberration coefficient as δ = λ3/4 Cs1/4

(6.8.5)

The spherical aberration coefficient, Cs , is particularly important in defining the performance of electromagnetic lenses used in electron microscopes and is discussed further in §9.2.2. Spherical aberration of the lens causes the image magnification to vary in proportion to the cube of the distance from the optic axis and gives rise to three

13 A paraxial ray makes a small angle (θ ) with the optical axis of the system, and lies close to the axis throughout the system.

OUP CORRECTED – FINAL

380

Optics, Optical Methods, and Microscopy principal distortions. This is illustrated in Figure 6.8.3c, for point objects displaced from the optic axis. If the magnification of the image increases with distance, it gives rise to pincushion distortion. Alternatively, if the image decreases with distance, a barrel distortion is observed. Finally, if the image is subject to an angular rotation as a function of its distance from the optic axis, it causes a spiral distortion and produces sigmoidal-shaped images. For further details, see Flewitt and Wild (2003). Finally, if the lens produces images without perfect axial symmetry, the image planes in two orthogonal directions differ from each other. In practice, this causes the vertical and horizontal components of the image to focus on different planes, and is known as astigmatism. As a result, no sharp image planes exist and there is a region of “confusion” between two sharply focused images. In optical systems, this defect is inherent in the manufacturing of the lens. 6.8.2.2

Depth of Field and Depth of Focus

The resolution possible for an object with its image in focus in the image plane, given by (6.8.2), depends on the numerical aperture, or nsinθ . Thus, the object can be displaced from its position without sacrificing the resolution (Fig. 6.8.4), and the distance it can be displaced along the optic axis and still remain in focus is called the depth of field, df . For a given resolution, δ, of the lens, it is given by df = δ tan α

(6.8.6)

where α is the half-angle subtended by the objective aperture at the focal point. The depth of field determines how stable the specimen stage should be for optimal imaging. If n sinα ∼ 1, then δ ∼ 0.6 λ ∼ 300 nm, and df = δ tan α 0.3 − 0.5 μm, which is the accuracy required for the specimen positioning system. Further, the distance over which the image remains in focus is called the depth of focus, D, and is given by D = M 2 df

df δ

(6.8.7)

Mδ

α

D

R u

v

Figure 6.8.4 For a finite resolution, the depth of field, df , gives the range of positions that the object can occupy and still produce a focused image. Similarly, the focused image may be observed over a range of distances called the depth of focus, D.

OUP CORRECTED – FINAL

Optical Microscopy 381 where M is the magnification, and df is the depth of field. Note that D is not so critical as it is dependent on the square of the magnification, and even for M = 100, it is of the order of a millimeter.

6.8.3

The Optical Microscope

An optical or light microscope is a versatile tool, regularly used in metallography (§6.11), for examining, evaluating, and quantifying the microstructure of materials. Optical microscopes operate in either transmission or reflection modes, and the images produced have a typical resolution of ∼250 nm. The reflection mode is particularly suited for examining opaque specimens of materials with the quality of the image enhanced by etching or polishing the surfaces of the specimens to be observed (see §6.11.1). Needless to say, care should be taken to ensure that the region of a specimen observed by any microscopy technique is truly representative of the sample. 6.8.3.1

Vertical Illumination in Reflection Geometry

A modern optical microscope (Fig. 6.8.5) has three components: the illuminating system, the imaging system, and the specimen stage which holds and moves the specimen in the X, Y, and Z directions, with mechanical stability consistent with the depth of field of the instrument. The imaging system consists of two lenses, one with very short focal length, called the objective, and the other, with a somewhat longer focal length, called the eyepiece (or ocular). The specimen or object (1) is placed just outside the focal point of the objective lens (Fig. 6.8.5b) and forms a real, magnified image at the first image plane (2), which becomes the object for the second lens, the eyepiece. The eyepiece magnifies (2) and forms a large virtual image (3), which then becomes the object for the eye to form a real final image (4) on the retina. The specimen is illuminated with the condenser system, which is adjusted to uniformly illuminate the objective aperture, and focus an image of the source on the back focal plane of the objective lens. By eliminating any unwanted light reflected from the numerous surfaces of the lens, the illumination is further optimized. The condenser aperture is used to limit the amount of light, with improved contrast obtained by using a smaller aperture diameter. The brightest available source of light should be used, e.g. a mercury arc lamp or tungsten-halide discharge tube (§5.2.1.1). Appropriate filters can be used if monochromatic light, such as green wavelength light, is desired for viewing. Example 6.8.3: You have two positive lenses, both with a focal length of 20 mm, to make an optical microscope (one of them serves as the objective and the other as the eyepiece). If the object is positioned 25 mm from the objective (a) what is the separation between the lenses, and (b) what is the expected magnification of this microscope. Assume that an object at the focal point forms an image at a distance of 25 cm.

OUP CORRECTED – FINAL

382

Optics, Optical Methods, and Microscopy (a)

(b)

Eye piece

(4) First image plane

Eye piece Virtual image aperture

Condenser aperture

(2)

Light source Half silvered mirror

Condenser lens

Objective back focal plane

Objective lens

Objective lens (3)

(1)

Specimen stage Figure 6.8.5 (a) An optical microscope and a schematic drawing of all its principal components in the reflection geometry and vertical illumination commonly used in metallography. (b) Ray diagram showing the image formation, including the eye of the viewer, in an optical microscope.

Solution: (a) The intermediate image distance can be calculated from the lens 1 1 formula (6.8.3), as 1f = 1u + 1v = 20 = 25 + 1v . Thus v = 100 mm. This is the distance from the objective lens to the intermediate image to which we need to add the focal length of eyepiece to get the separation of the two lenses: 100 mm + 20 mm = 120 mm. (b) The magnification of the objective is Mobj = v/u = 100/25 = 4× (inverted image) For the eyepiece, the object is at the focal point and the image is formed at a distance of 25 cm. Thus, the magnification of the eyepiece is Mep = 250/20 = 12.5×. The total magnification of the microscope M = Mobj × Mep = 4 × 12.5 = 50×.

OUP CORRECTED – FINAL

Optical Microscopy 383 6.8.3.2

Direct, Oblique, and Dark Field Imaging

In addition to direct illumination discussed previously, which produces a bright field image, one can obtain an oblique or dark field image (Fig. 6.8.6a) by tilting the incidence angle. Oblique incidence can be obtained using off-axis illumination by simply shifting the condenser aperture (Fig. 6.8.6b). In dark field imaging, light incidence is such that no specularly reflected light enters the objective lens; in practice, instead of tilting the illumination, this is achieved by using an annular cone of light focused on the object plane. Dark field illumination is limited by low-image brightness but with an improvement in contrast (Fig. 6.8.6c). 6.8.3.3

Interference Contrast Microscopy

In its simplest form (Fig. 6.8.7a), a half-silvered cover slip is placed above the specimen, which splits the beam into two, half of which is reflected back from the cover slip, ray 1, and the other half, ray 2, is reflected from the specimen surface. Increasing tilt of vertical illuminator (b)

Oblique

(c) Dark Field

(d)

(e)

Intensity

(a) Bright Field

Distance

50μm

Figure 6.8.6 (a) Direct or bright field, (b) oblique, and (c) dark field illumination of a specimen, obtained by progressively tilting the illuminating beam. The reflected intensity distribution for each of these cases is also shown. Slip traces observed (d) in deformed β-phase by oblique incidence (note dark contrast), and (e) in α-phase (Bainite) by dark field imaging (note bright contrast), in a Cu–40%Zn alloy. Adapted from Flewitt and Wild (2003).

(a)

(1)

(b) (2) Phase difference φ = 2πh / λ

Half-silvered Coverslip h Specimen

Figure 6.8.7 (a) Schematic representation of the two-beam interference method, and (b) image of a groove along a grain boundary. Adapted from Brandon and Kaplan (2008).

OUP CORRECTED – FINAL

384

Optics, Optical Methods, and Microscopy The two beams, which were originally coherent, now have a phase difference depending on the additional path length ray 2 has to travel to the specimen surface and back. This phase difference leads to interference fringes (Fig. 6.8.7b) that can be interpreted in terms of local thickness variations. Assume that the coefficient of reflection for the half-mirrored glass slip is R; then (1−R) is the transmission coefficient. After reflecting from the specimen surface, again only (1−R) of the intensity is transmitted by the slip; as a result (1−R)2 of the original intensity is reflected from the specimen. For strong interference, the original reflected intensity, R, should be the same as the final transmitted intensity, (1−R)2 . Equating terms, we get R = (1 − R)2 , or R =  √  1/ 1 + 2 = 1/2.414 0.41. For the total path difference, 2h, of the two beams to give rise to destructive interference (black fringe), it must be equal to (2n + 1) λ/2, where n is an integer and λ is the wavelength. Thus, when (2n + 1) λ/2 = 2h is satisfied, dark fringes occur for height differences, h, of λ/4, 3λ/4, 5λ/4.... and successive fringes reflect height contours, h = λ/2. To detect local variations in specimen height or topography we need to detect changes in the positions of the fringes. If we assume that 10% shift in fringe separation can be detected, and using green light (λ ∼ 550 nm), we can expect to resolve height differences with an accuracy of ±25–30 nm. Further improvement in such twobeam interference microscopy can be achieved by using a drop of immersion oil between the coverslip and the specimen to effectively change the wavelength by the factor, n, the refractive index of the oil. Interference contrast microscopy finds particular use in detecting details from optically transparent specimens that would otherwise lead to minimum contrast in a traditional optical microscope. 6.8.3.4

Optical Microscopy with Polarized Light

As seen in §1.3.4, a polarized light beam has its electric field vector component, E, aligned in a specific direction normal to the direction of propagation. Recall that when the direction of E is fixed in one plane but its magnitude varies along the propagation direction (Fig. 6.2.7) the light is called plane polarized. Further, following our discussion in §6.2.6, it is easy to see that if two plane polarized waves with their vectors inclined at some angle to one another are combined, it results in a third plane polarized wave with a vector sum of the original waves. Alternatively, a plane polarized wave can also be resolved into two orthogonal components, lying on arbitrary planes separated by 90◦ (Fig. 6.8.8b). In a polarizing microscope, Figure 6.8.9a, the incident light is plane polarized by inserting a polarizer into the path of the condenser system. When this plane polarized light is reflected from an optically anisotropic surface, it undergoes a phase change and becomes elliptically polarized (see §6.2.6) and the electric vector now has a component orthogonal to the plane of polarization of the initial light. A second polarizer, called the analyzer, is placed between the objective lens and the eyepiece but with its polarization plane at 90◦ to the first polarizer; this

OUP CORRECTED – FINAL

Optical Microscopy 385 (a)

(b)

y

Plane of polarizer

(1) (2)

(2)

z

Principle axes of surface anisotropy

x Resolved components

(1)

Incident wave vector Figure 6.8.8 (a) Plane polarized light consisting of two beams in phase and polarized orthogonal to one another. (b) A plane polarized light beam can be resolved into arbitrary orthogonal components. Adapted from Brandon and Kaplan (2008).

(a)

Plane of polarizer

(b)

Incident amplitude

Eyepiece First image Plane polarised incident beam Analyzer Elliptically polarised reflected beam

Polarizer

Amplitude accepted by analyzer

Resolved components reflected from the sample Plane of analyzer

Light source Half silvered mirror Condenser lens

(c)

Plane polarised incident beam Objective back focal plane Objective lens

Specimen 100 μm Figure 6.8.9 (a) An optical microscope with crossed-polarizer and analyzer used for imaging. (b) A plane polarized beam reflected from an optically anisotropic specimen surface will be elliptically polarized. An analyzer, polarized orthogonal to the analyzer, picks up the perpendicular component to form the image. (c) Details of the crystallization process of a polymer revealed by polarizing microscopy. Adapted from Brandon and Kaplan (2008).

OUP CORRECTED – FINAL

386

Optics, Optical Methods, and Microscopy will detect the vector component perpendicular to the plane of polarization of the incident light (Fig. 6.8.9b). Thus, the phase change at the reflecting surface can be measured and specific features can be imaged, e.g. the crystallization process of polymers (Fig. 6.8.9c).

6.8.4 Confocal Scanning Optical Microscopy (CSOM) CSOM, in which a finely focused beam of light is rastered over the specimen surface, is the light analog of scanning electron microscopy (§10). It is widely used in biological imaging. In conventional optical microscopy, discussed thus far, the entire field of view of the specimen is imaged. Moreover, the signal is also collected above and below the focal plane, which contributes significant blur to the final image degrading its contrast and sharpness (Fig. 6.8.11). In contrast, in CSOM, the illumination of the specimen is restricted to a single point at a specific depth of the specimen, which is then scanned to produce a complete image. Further, a confocal imaging aperture is introduced in the optical pathway, such that all light coming from regions above and below the focal plane of the microscope is attenuated and does not contribute to the image. Thus, the final image only contains the in-focus information. Figure 6.8.10 shows the principle of confocal microscopy. The laser light from the source passing through the illuminating aperture, is reflected by the dichroic

Photomultiplier Confocal aperture

Illumination aperture

Dichroic mirror

In-focus rays Out-of--focus rays

Point source (laser)

Figure 6.8.10 Schematic representation of a confocal microscope (see text for details).

Objective lens

Adapted from Sheppard and Shotton (1997).

Focal plane

Specimen

OUP CORRECTED – FINAL

Optical Microscopy 387 100 μm

(a) 100 μm

(c)

100 μm

(b) 100 μm

(d)

mirror,14 and then focused by the microscope objective lens to a diffraction limited spot at the focal plane of a three-dimensional specimen. Reflected or fluorescent signal is then produced from an imaging volume (voxel) at the focal depth, as well as from regions above and below it. However, only light from the in-focus voxel is allowed to pass through the confocal aperture to be detected; all other contributions from regions different from the focal plane are severely attenuated and do not contribute anything to the final image. Note that unlike conventional optical microscopes, which have a limited depth of field (§6.8.3.2), a CSOM can be operated as though it had an infinite depth of field by slowly and progressively changing its focal plane in depth for each scan of the specimen surface. Figure 6.8.11a is a CSOM image, taken in reflection mode, of a small region of a tilted microcircuit using light of 633 nm wavelength. It is compared with a standard optical image (Fig. 6.8.11b) that shows significant blur, and a CSOM image (Fig. 6.8.11c) of the entire specimen showing all of it in focus. CSOM combined with internal variations in autofluorescence is a versatile technique of particular importance in biology and the life sciences. Figure 6.8.12b illustrates this method of imaging through a sunflower pollen grain with optical sections gathered in 0.5 μm steps along the microscope optical axis. Pollen grains (20–40 μm in diameter) typically yield blurred images in a conventional fluorescence microscope (Fig. 6.8.12a); needless to say, they also lack information about internal structural details.

Figure 6.8.11 Comparison of confocal scanning optical microscope image of an electronic microcircuit with that obtained from a conventional full-field image. (a) In CSOM, a small region where only the central infocus region is imaged, with all other regions appearing dark. (b) Conventional optical microscopy showing significant blur in the image. (c) The entire CSOM image of the specimen, all of it in focus. (d) Same as (c) but without the confocal optics; notice the significant blurring. Adapted from [1].

14 A dichroic mirror shows significantly different reflection or transmission properties at two different wavelengths.

OUP CORRECTED – FINAL

388

Optics, Optical Methods, and Microscopy (a)

(b)

1

2

3

4

5

6

7

8

9

10

11

12

(c)

Figure 6.8.12 Images of a sunflower pollen grain using (a) conventional optical microscopy, and (b) CSOM sections (1–12) starting from the top in 0.5 μm steps along the optic axis using a dual argon-ion (488 nm; green fluorescence) and green helium/neon (543 nm; red fluorescence) laser system, showing its internal structure. (c) A false color scanning electron microscope (§10) image of the same pollens. (a–b) Downloaded with permission from http://olympus.magnet.fsu.edu/primer/techniques/confocal/confocalintro.html. (c) Courtesy Marie Curie Institute.

In summary, CSOM excludes most of the light from the specimen that is not from the focal plane of the microscope. Thus, the image has better contrast than that of a conventional microscope and allows not only better observation of fine details but also helps build three-dimensional reconstructions of the specimen by assembling a series of images of thin slices taken along the optical axis. The excellent handbook by Sheppard and Shotton (1997) provides further details of CSOM.

6.8.5

Metallography

The simplicity and accessibility to optical microscopy makes it a widely used method to evaluate microstructures of materials. Some practical comments on specimen preparation and the application of optical microscopy are included here;

OUP CORRECTED – FINAL

Optical Microscopy 389 (c)

(a)

Nickel (b)

(e)

Low Carbon Steel (d)

Bronze Alloy (f)

Figure 6.8.13 Effect of etching in metallography. The control of the degree of etching is important as shown for a 270-grade nickel etched with Kallings #2 reagent for (a) under-etched and (b) properly etched specimen under bright field imaging. Different etchants reveal different aspects of the microstructure shown here in bright field for a low-carbon steel specimen etched with (c) 2% Nital and (d) 4% Picral. Similarly, grain structures of a phosphor bronze alloy etched by (e) equal parts NH4 OH and 3% H2 O2 , and (f) Klemm’s I tint etch. All figures adapted from Williams, Pelton, and Gronsky (1991).

further details on metallography can be found in specialized monographs (Vander Voort, 1984). In the examination of materials by optical microscopy, the preparation of specimens for observation is critically important and involves appropriate sectioning and polishing of their surface, followed by etching. For best results, it is best to examine specimens in the as-polished condition as it reveals certain microstructural features, such as inclusions, intermetallic phases, cracks, or porosity, which may be obscured by etching. Metal specimens are often etched with acid or base solutions to reveal the details of the microstructure. Initially, a general-purpose etchant (Table 6.8.1) is used, and is then followed by more specialized ones (Fig. 6.8.13). Metallography is largely used to study opaque specimens; thus bright-field imaging in reflection mode is often used. However, optically anisotropic15 specimens are best examined by cross-polarized illumination (Fig. 6.8.14a). Moreover, for examination with cross-polarized illumination, dark field illumination (Fig. 6.8.14b), or interference contrast imaging (§6.8.3.3), the specimens must be polished to a higher quality than what is required for bright field illumination to avoid seeing fine scratches in the images.

15 Interacts with light differently in different crystallographic directions.

OUP CORRECTED – FINAL

Table 6.8.1 Etchants commonly used in metallography. Name of Etchant

Composition

Conditions for use

Suitable for etching

ASTM #30

Ammonia – 50 ml Hydrogen Peroxide (3%) – 100 ml DI water – 50 ml

Copper, copper alloys, and Cu–Si alloys.

Adler Etchant

Copper ammonium chloride – 6 g Hydrochloric acid – 100 ml Ferric chloride, hydrated – 30 g DI water – 50 ml. FeCl3 – 7 g CuCl2 – 2 g Hydrochloric acid – 100 ml Nitric acid – 4.9 ml Ethanol – 100 ml CuCl2 –5 g, Hydrochloric acid – 100 ml, Ethanol – 100 ml Hydrochloric acid – 3 ml Nitric acid – 5 ml Hydrofluoric acid – 2 ml DI water – 190 ml.

Mix ammonia and water before adding peroxide. Use fresh. Swab surface for 5–45 s. Combine all, immerse specimen for several seconds.

Carpenter’s stainless steel etch

Kalling’s #2

Keller’s etch

300 series stainless steels, Hastelloy, superalloys.

Combine all, immerse specimen for several seconds at 20 ◦ C.

Duplex and 300 series stainless steels.

Combine all, immerse or swab specimen at 20 ◦ C.

Duplex and 400 series stainless steel, Ni–Cu alloys and superalloys. Al and Ti alloys.

Always use fresh etchant. Immerse specimen for 10–30 s.

Klemm’s reagent

Sodium thiosulfate, saturated – 250 ml Potassium metabisulfite – 5g.

Etch for few seconds to minutes.

Kroll’s reagent

Nitric acid – 6 ml Hydrofluoric acid – 2 ml DI water – 190 ml. Nitric acid – 1–10 ml Ethanol – 100 ml.

Swab specimen for maximum 20 s.

Nital

Picral

Picric acid – 2–4 g Ethanol – 100 ml.

Immerse specimen for up to 1 min. Seconds to minutes. Prevent etchant from drying or crystallizing as it can explode!

α–β brass, bronze, tin, cast-iron phosphides, ferrite, martensite and retained austenite. Ti and its alloys.

Iron, carbon and alloy steels, cast iron. Mn- alloys, magnetic alloys, Mg. Microstructures containing ferrite, carbide, pearlite, martensite, and bainite.

OUP CORRECTED – FINAL

OUP CORRECTED – FINAL

392

Optics, Optical Methods, and Microscopy (a)

(b)

25 μm

25 μm

(c)

Figure 6.8.14 Metallography of annealed ductile iron containing graphite nodules in (a) bright-field at 400×, and (b) cross-polarized light microscopy. MnS with calcium– manganese sulfide tips shown in (c) bright field, and (d) dark field imaging.

5 μm (d)

5 μm

All figures adapted from Williams, Pelton, and Gronsky (1991).

6.9 Ellipsometry Ellipsometry is an optical measurement technique that is used to determine properties of specimens, largely thin films, based on changes in the polarization state of light reflected (transmitted) from (through) the specimen surface. The polarization change is measured in terms of an amplitude ratio, , and a phase change, (§6.9.3). The technique is primarily used to determine optical constants and film thickness, but with additional sophistication in analysis it can also measure chemical composition, roughness, and crystallinity (see Fujiwara, 2007). In general, the interaction of light with matter can be described in terms of a complex refractive index, N, which consists of the index of refraction, n, describing the phase velocity of the light in the medium as v = c/n, and the material dependent extinction coefficient, K, such that

OUP CORRECTED – FINAL

Ellipsometry 393 N = n + iK

(6.9.1)

Further, the optical properties of a material are described by the complex dielectric constant ε = ε1 + iε2

(6.9.2)

where ε = N 2 , ε1 = n2 − K 2 and ε2 = 2nK. Even though the light slows down as it enters a medium with higher index, its frequency remains unchanged, and its wavelength is shortened. The extinction coefficient, K, describing the loss of energy from the light to the medium, is given by μ = 4π K/λ

(6.9.3)

where μ describes the attenuation of light (Beer’s law) in the medium of thickness, t, introduced earlier, §1.3.5.1, as (1.3.11): I = I0 exp (−μt). This is also illustrated in Figure 6.9.1; however, when there is no absorption in the medium, μ = K = 0. Furthermore, the complex dielectric constant and the complex refractive index are related through

n=

 1/2 1/2 ε1 + ε12 + ε22 2

(6.9.4)

(a) K=0 E0

Et0 x

λ (b)

K>0

λ/n

E t0 exp(–2πKx/λ) E0 x

λ

λ/n

Figure 6.9.1 The propagation of an electromagnetic wave (only Ecomponent shown) in (a) a transparent medium (K = 0), and (b) in an absorbing medium (K > 0). Note that in the latter, E 0 > E t0 , as some of the light is reflected at the interface. Adapted from Fujiwara (2007).

OUP CORRECTED – FINAL

394

Optics, Optical Methods, and Microscopy

K=

1/2 1/2  −ε1 + ε12 + ε22 2

(6.9.5)

Finally, from K, (6.9.4), and using (6.9.3) the absorption coefficient, μ, can also be obtained.

6.9.1 p- and s-Polarized Light Waves, and Fresnel Equations of Reflection In ellipsometry, we are interested in the reflection of polarized light at oblique incidence from a specimen surface (Fig. 6.9.2). It is common practice to classify the light into s- and p-polarization components, depending on the direction of oscillation of its electric field. In p-polarization (Fig. 6.9.2), the electric field of the light oscillates in the plane of incidence and reflection, whereas in s-polarization, the electric field oscillates in a plane orthogonal to this plane of incidence and reflection. Figure 6.9.3a,b shows the propagation of the electric field, E, and magnetic induction, B, of p- and s-polarized light upon reflection from a surface. The ability of a surface to reflect a specific electromagnetic wave is characterized by the amplitude reflection coefficient. In general, upon reflection both E and B components of the light wave have to satisfy the boundary condition that their components parallel to the interface are continuous at the interface. Thus, the parallel components on the incidence side (throughout this section, we use subscripts i for incidence, r for reflection, and t for transmission) should be equal to that on the transmitted side. Hence, for the p-polarized light (Fig. 6.9.3a) we get Eip cos θi − Erp cos θr = Etp cos θt p s Plane of incidence

Figure 6.9.2 Reflection of s- and ppolarized light waves from a surface. Adapted from Fujiwara (2007).

(6.9.6a)

OUP CORRECTED – FINAL

Ellipsometry 395 (a)

Erp

Eip

(b)

–Erp cosθr

Eip cosθi

Brp

Bip ni

Bis

θi θr

nt

Ers

Eis

–Bis cosθi

θi

ni nt

θt

Brs

θt

Etp Btp

θr

Brs cosθr

Ets

–Bts cosθt Etp cosθt

Bts

p–polarization

s–polarization

Figure 6.9.3 The electric field, E, and magnetic induction, B, for (a) p-polarization and (b) s-polarization, at the interface between two different materials. Adapted from Fujiwara (2007).

Bip + Brp = Btp

(6.9.6b)

Further, for a medium with refractive index, n, E = vB = Bc/n. Hence, (6.9.6b) can be rewritten as   ni Eip + Erp = nt Etp

(6.9.6c)

By eliminating Etp between (6.9.6a) and (6.9.6c), and setting θi = θr , we can obtain the amplitude reflection coefficient, rp , for p-polarization rp =

Erp nt cos θi − ni cos θt = Eip nt cos θi + ni cos θt

(6.9.7)

Similarly, for s-polarized light (Fig. 6.9.3b) the boundary condition at the interface gives Eis + Ers = Ets

(6.9.8a)

−Bis cos θi + Brs cos θr = −Bts cos θt

(6.9.8b)

which leads to amplitude reflection coefficient rs =

Ers ni cos θi − nt cos θt = Eis ni cos θi + nt cos θt

(6.9.9)

OUP CORRECTED – FINAL

396

Optics, Optical Methods, and Microscopy The Fresnel equations for reflection (6.9.7; 6.9.9) are valid even if the refractive index, n, is replaced by the complex refractive index, N, where ε = N 2 . It can then be shown that rp =

εt Nii − εi Ntt εt Nii + εi Ntt

(6.9.10)

rs =

Nii − Ntt Nii + Ntt

(6.9.11)

 1/2 . This can also be written in where Nii = Ni cos θi and Ntt = εt − εi sin2 θi polar coordinates as   rp = | rp | exp iδrp

(6.9.12)

rs = | rs | exp(iδrs )

(6.9.13)

In practice, in ellipsometry we measure the ratio of the reflection amplitude coefficients, rp /rs . Without proof, it is worth mentioning that the difference between rp and rs is maximized, to first order, for the angle of incidence equal to the Brewster angle, θB , defined as tan θB = nt /ni

(6.9.14)

At the air/glass interface, nt /ni = 1.49 and θB = tan−1 (1.49) = 56◦ . Further, θB depends on the wavelength, and for typical semiconductors, measurements are carried out at θi ∼ 60-80◦ . Example 6.9.1: Fused glass has a refractive index, n = 1.45. For a film of this material placed in air and with Brewster angle incidence, calculate the reflection coefficient for both s- and p-polarized light. Solution: The Brewster angle, (6.9.14), is θB = tan−1 (1.45) = 55.4◦ = θi . The angle of refraction is given by Snell law, (6.5.2), θt = sin Then, rp = And rs =

−1

Erp Eip

Ers Eis

=



=

ni sinθi nt



−1

= sin

nt cos θi − ni cos θt nt cos θi + ni cos θt

ni cos θi − nt cos θt ni cos θi + nt cos θt

=

=



1.0 sin55.4◦ 1.45



= 34.6◦ .

1.45 cos(55.4) − 1.0 cos(34.6) 1.45 cos(55.4) + 1.0 cos(34.6)

cos(55.4) − 1.45 cos(34.6) cos(55.4) + 1.45 cos(34.6)

(6.9.14)

= 0.0001.

= −0.3552.

OUP CORRECTED – FINAL

Ellipsometry 397

6.9.2

Optical Elements Used in Ellipsometry

To determine the polarization state of the light, optical components used in ellipsometry include polarizers (analyzers) and retarders. Typically, the former is used to extract linearly polarized light from unpolarized light, and the latter are used to convert linearly polarized light to circularly polarized light. Figure 6.2.6 shows three cases of the polarization state—linear, circular, and elliptical— typically encountered in ellipsometry as a function of the phase difference,

= δx − δy , assuming the same amplitude, Ex0 = Ey0 for the two phase components. Recall from §4, Table 4.1.1, that a cubic crystal, such as NaCl, has its atoms arranged in a highly symmetric form with four threefold axes along the directions. Light emanating from a point source within the material will propagate uniformly in all directions. Thus, it will be characterized by a single index of refraction. Alternatively, if the crystal belongs to the hexagonal, tetragonal, and trigonal systems, with uniaxial symmetry, the atoms are so arranged in the unit cell that the propagating light will encounter a different or asymmetric interaction depending on the direction of propagation. These materials are optically anisotropic and birefringent,16 and there is only one direction (optic axis) about which the atoms are symmetrically arranged. Thus, these materials have two principal indices of refraction, n⊥ and n , perpendicular and parallel to the optic axis, respectively. The difference, n = n − n⊥ , is a measure of the birefringence. Calcite is a good example of a uniaxial, optically active crystal. Finally, crystal systems of even lower symmetry (orthorhombic, monoclinic, and triclinic) have two optic axes and are called biaxial. Such crystals have three principal indices of refraction. In practice, the birefringence of a biaxial crystal (e.g. mica) is the difference between its largest and smallest values of the indices of refraction. 6.9.2.1

Polarizer (Analyzer)

A standard polarizer is made of two prisms of a uniaxial calcite crystal (Fig. 6.9.4a). Such a polarizer (analyzer) is also known as a Glan–Taylor prism, and produces linearly polarized light along the optical anisotropy axis of the crystal. 6.9.2.2

Compensator (Retarder) and Photoelastic Modulator

In a typical ellipsometry set up, a compensator is placed either in front of the analyzer or after the polarizer. Its main function is to convert linear to circularly polarized light, and uses an anisotropic birefringent crystal that has two orthogonal axes along which light propagates with different velocities. The propagating light, initially linearly polarized at 45◦ , is then converted to left circularly polarization (Fig. 6.9.4b). For a thickness, d, of the compensator, this is accomplished by generating a phase difference, δ, between Ex and Ey , due to differences in their velocities of propagation, where δ=

2π (ne − no )d λ

(6.9.15)

16 Having indices.

two

different

refractive

OUP CORRECTED – FINAL

398

Optics, Optical Methods, and Microscopy Ey

(a)

(b)

Slow axis

Ey

Ex

Ey

Ex Ey Fast axis

Ex : transmission axis

Ex

Ex

A Glan–Taylor prism

d Figure 6.9.4 (a) A Glan–Taylor prism used as a polarizer (shown) or as an analyzer. (b) A compensator converts linearly polarized light to circularly polarized light by using a birefringent crystal that causes light to propagate with different velocities in orthogonal directions. Adapted from Fujiwara (2007).

and ne and no are the indices of refraction for the two components traveling with different velocities. Needless to say, δ depends on both thickness and wavelength, and for the special case of δ = π/2, which corresponds to a wavelength difference,

λ = λ/4, the compensator is known as a quarter wave plate. In many isotropic materials, the application of stress leads to optically anisotropic behavior, called photoelasticity, and introduces birefringence behavior proportional to the applied stress and with optic axis aligned along the direction of stress. The phase shift, δ, in such photoelastic modulators varies continuously with time; details are beyond the scope of this book and can be found in more advanced texts (Fujiwara, 2007).

6.9.3

Ellipsometry Measurements

As mentioned, ellipsometry measurements monitor the change in polarization of p- and s-polarized light, Eip and Eis , optimally incident at the Brewster angle and reflected from the specimen, to obtain optical constants and film thickness. Figure 6.9.5 shows a typical measurement set up. Note that when θ = 0◦ , the incident and reflected waves overlap completely; also, for the incident light (shown) linearly polarized at 45◦ , Eip = Eis is satisfied. As mentioned (§6.9.1), the amplitudes of the reflected light for p- and s-polarization differ significantly, and ellipsometry measures their amplitude ratio, , and the phase difference, , between the s- and p-polarization. In the simplest case,  is characterized by the refractive index, n, and the phase difference, , represents the light absorption as described by the extinction coefficient, K. In other words, (n,K) can be determined from (, ) by applying the Fresnel equations (6.9.7; 6.9.9). Figure 6.9.6 shows the measured ellipsometry spectra of SiO2 , for two different thickness, illustrating the variation of the amplitude ratio, , and phase lag, , as a function of wave number.

OUP CORRECTED – FINAL

Ellipsometry 399 Specimen s E

n, K θ

p

Eis

s p

Eip

Ers



Erp ψ

Figure 6.9.5 Top: Measurement principle of ellipsometry. The amplitude ratio, , and the phase lag,

, are measured and interpreted in terms of the optical constant, n and K, and other parameters of the thin film. Bottom: A commercial ellipsometer used for routine analysis of thin films. Top: Adapted from Fujiwara (2007). Bottom: Courtesy University of Washington.

50

250 130 nm

484 nm 150

30 50 20 –50

10 0 1000

(deg)

(deg)

40

1200 1100 Wave number (cm–1)

1300

Figure 6.9.6 Ellipsometry spectra ( and ) for two different thickness (130 nm and 484 nm) of SiO2 , as a function of wavenumber.

–150

Adapted from [3].

In ellipsometry, the amplitude ratio, ρ, is related to  and by the following relationship ρ=

rp = tanexp(i ) rs

(6.9.16)

Now, using (6.9.9) and (6.9.7) we can rewrite (6.9.16) as ρ=

rp Erp /Eip = tanexp(i ) = rs Ers /Eis

(6.9.17)

In a simple case (Fig. 6.9.5), Eip = Eis , and hence (6.9.17) is modified as ρ= Further, in polar coordinates

Erp rp = tanexp(i ) = rs Ers

(6.9.18)

OUP CORRECTED – FINAL

400

Optics, Optical Methods, and Microscopy tan =

| rp | and = δrp − δrs . | rs |

(6.9.19)

In summary, ellipsometry measures the ratio of two values, and as such, is very robust, accurate, and reproducible. It is a model-based method, and after  and are experimentally determined, a model of the thin film is built, unknown optical constants and/or thickness parameters are varied, and  and values are calculated using the Fresnel equations (6.9.7) and (6.9.9). The calculated  and values, which best match the experimental ellipsometry data provide the optical constants and thickness parameters of the specimen. Depending on the material, thickness in the range of a few Å to a few micrometers can be measured by ellipsometry; however, for optically opaque materials, such as metals with strong absorption and shallow penetration of light, ellipsometry can routinely measure only optical constants, and not the thickness [2].

Summary

17 Jean-Robert Argand (1768–1822) was a Swiss/French amateur mathematician.

Propagation of light can be described as the simple harmonic motion of transverse waves, where the rate of change of its phase with time (distance) gives its angular frequency (wave vector). The linear superposition, or addition, of two or more SHWs produces a new SHW with a different amplitude and phase angle that can be derived from a simple geometric construction known as the phasor representation. Alternatively, it can be solved for analytically using complex notation and the related Argand17 diagram. Additionally, combining waves that propagate on orthogonal planes can give rise to linear, elliptical, or spherical polarization, depending on their amplitudes and phase differences. Two classical experiments inform our understanding of the behavior of electromagnetic waves. Huygens studied the propagation of light through a pin hole and, to explain the observed intensity distribution (which deviated significantly from geometric optics), he postulated that every point on an advancing wave front serves as a source of new spherical wavelets. The subsequent Young doubleslit experiment demonstrated the principle of optical interference and diffraction. These experiments paved the way for our understanding of diffraction, which is now broadly classified as Fraunhofer and Fresnel diffraction. Fraunhofer diffraction is expected when the incident and diffracted waves are plane waves, i.e. the source and detector are significantly far away that the curvature of the wave front can be neglected, and is relevant to the understanding of the behavior and resolving power of diffraction gratings (§3.5.1) used in a wide range of spectroscopy measurements. Moreover, generalization of Fraunhofer diffraction to scattering by a three-dimensional arrangement of scatterers (Figure 1.4.7), such as atoms in crystals, forms the basis of diffraction methods (X-rays, §7, and electrons and neutrons, §8) in materials characterization. In contrast, Fresnel diffraction is observed when either the source or detector is close enough that the

OUP CORRECTED – FINAL

Further Reading 401 curvatures of the wave front becomes significant. From a practical point of view, Fresnel diffraction finds application in the design of zone plates used in focusing high-energy X-rays for microscopy at synchrotron radiation facilities. Optical microscopy relies on visual observation and works best when the instrumentation is tailored to the characteristics of the human eye. The latter is sensitive to the visible spectrum (∼380 nm < λ < 700 nm), but with peak sensitivity for green light (λ = 550 nm). The Rayleigh criterion defines the resolution for any imaging system. For the human eye this resolution is ∼0.1mm at the optimal wavelength. Further, matching the resolution of the eye with that of an optical microscope, it can be shown that there is an upper limit (500×) for the magnification of the microscope. Lenses, in general, used in all microscopy suffer from three major defects—spherical and chromatic aberrations, and astigmatism—that limit their performance. Optical microscopes work in transmission or reflection modes; the latter is particularly useful for studying opaque specimens. In addition to direct illumination producing bright field images, microscopes can be operated under oblique and dark field imaging modes, or to produce interference contrast, or imaging with polarized light. A newer variation of optical imaging is confocal scanning microscopy, where the light (probe) is focused to a single point at a specific depth in the specimen, and then scanned to produce a complete image. Metallography, widely used to examine materials by optical microscopy, provides a wealth of information on the materials microstructure. It requires that the surfaces be mechanically polished, and if necessary, chemically etched to provide optimal contrast. Finally, the polarization state of light reflected (or transmitted) from the surface of a specimen can be studied to obtain details of the optical properties of the material. Known as ellipsometry, this technique finds wide use in the characterization of materials, particularly thin films. .............................................................................. FURTHER READING

Attwood, D. Soft X-Rays and Extreme Ultraviolet Radiation. Cambridge: Cambridge University Press, 2000. Brandon, D., and W. D. Kaplan. Microstructural Characterization of Materials, 2nd ed. Chichester: Wiley, 2008. Flewitt, P. E. J., and R. K. Wild. Physical Methods for Materials Characterization. Boca Raton: IoP Press, 2003. Fowles, G. R. Introduction to Modern Optics, 2nd ed. New York: Dover, 1989. Fujiwara, H. Spectroscopic Ellipsometry: Principles and Applications. Hoboken: John Wiley & Sons, 2007. Hecht, E. Optics. Reading: Addison Wesley, 2001. Jenkins, F. A., and H. E. White. Fundamentals of Optics. Auckland: McGraw-Hill, 1981.

OUP CORRECTED – FINAL

402 Optics, Optical Methods, and Microscopy Sheppard, C. J. R., and D. M. Shotton. Confocal Laser Scanning Microscopy. Oxford: BIOS Scientific Publishers, 1997. Vander Voort, G. F. Metallography: Principles and Practice. New York: McGrawHill, 1984. Williams, D. B., A. R. Pelton, and R. Gronsky, eds. Images of Materials. Oxford: Oxford University Press, 1991. .............................................................................. REFERENCES

[1] Wilson, T., and D. K. Hamilton. “Dynamic Focusing in the Confocal Microscope.” Journal of Microscopy 128, no. 2 (1982): 139–43. [2] Rothen, A. “The Ellipsometer, an Apparatus to Measure Thicknesses of Thin Surface Films.” Review of Scientific Instruments 16 (1945): 26. [3] Downloaded from OSA Publishing (2019). [4] Paterson, S. M., Y. S. Casadio, D. H. Brown, J. A. Shaw, T. V. Chirila, and M. V. Baker. “Laser Scanning Confocal Microscopy Versus Scanning Electron Microscopy for Characterization of Polymer Morphology: Sample Preparation Drastically Distorts Morphologies of Poly(2-Hydroxyethyl Methacrylate)-Based Hydrogels.” Journal of Applied Polymer Science 127, no. 4 (2013): 4296–4304. .............................................................................. EXERCISES

A. Test Your Knowledge Identify ALL the correct statements for each statement. 1. All light waves (i) vibrate along the direction of propagation. (ii) vibrate in a plane normal to the direction of propagation. (iii) are transverse waves. 2. The addition of any two waves with displacements in orthogonal planes but propagating in the same direction (i) always gives rise to polarized light. (ii) gives rise to linearly polarized light if their phase difference is 0◦ . (iii) gives rise to circularly polarized light if their phase difference is π/2. 3. Huygens principle states that (i) light travels in straight lines. (ii) light is bent when it hits an aperture. (iii) every point on the advancing wave front of light is a new source of light waves.

OUP CORRECTED – FINAL

Exercises 403 4. In the Young’s double slit experiment, we observe (i) an interference pattern. (ii) a pattern that depends on the phase difference between two waves. (iii) intensity maxima that satisfy mλ = d sin θ. 5. The refractive index of a medium (i) is the ratio of the speed of light in vacuum to the speed of light in the medium. (ii) depends on the wavelength of light in the medium. (iii) can cause total internal reflection. 6. (i) Fresnel and Fraunhofer diffraction are the same. (ii) In Fresnel diffraction the source of light is at a finite distance. (iii) In Fraunhofer diffraction BOTH the source of light and the screen are at infinite distance. (i) In Fresnel diffraction a spherical wave front is incident on the object. (ii) In Fraunhofer diffraction a uniform plane wave is incident on the object. 7. For Fraunhofer diffraction through a slit (i) the spacings of the minima are proportional to the size of the slit. (ii) the spacings of the minima are inversely proportional to the size of the slit. (iii) the spacing of the observed minima in two perpendicular directions are the same for a rectangular slit. 8. An Airy disc (i) is light and floats in air. (ii) is the bright central disc observed in Fraunhofer diffraction when a _____________ aperture is used. (iii) is always observed in Fresnel diffraction. 9. A zone plate (i) is used to focus X-rays. (ii) If used to focus X-rays, satisfy the object and image distance given by the standard lens formula, 1a + 1b = 1f . (iii) in practice is made of alternating rings of W and Si. 10. The human eye (i) is uniformly sensitive to all wavelengths of visible light. (ii) has a peak sensitivity to green light. (iii) has a variable integration time. (iv) has an integration time of 0.1 s. (v) when dark adapted can form images with 100 photons/pixel. 11. The light microscope to be used by human eyes (i) should have the highest magnification possible. (ii) should have an optimal magnification of 300–400× (iii) with a magnification, M > 1000, would not be particularly useful.

OUP CORRECTED – FINAL

404 Optics, Optical Methods, and Microscopy 12. The best lenses have (i) no defects. (ii) chromatic and spherical aberrations. (iii) a depth of field and a depth of focus that are of the same magnitude. 13. The optical microscope (i) is a versatile tool for materials research and development. (ii) is regularly used in metallography. (iii) has only two components: the illumination and imaging systems. (iv) has only vertical illumination. 14. The optical microscope can produce images with (i) oblique incidence. (ii) interference contrast. (iii) polarized light. 15. Confocal microscopy (i) is the optical analog of an SEM. (ii) has an infinite depth of field. (iii) causes significant blurring of images. 16. In metallography (i) the quality of images can be improved by suitable etching of the specimen. (ii) the quality of images can be improved by polishing, especially for dark field illumination. (iii) it is best to first examine the specimen after polishing but before etching. 17. Ellipsometry (i) is an optical method to determine optical properties of materials. (ii) measures changes in the polarization of light upon reflection from a surface. (iii) involves s- and p-polarized light. (iv) when necessary, uses a half-wave plate. (v) is a curve fitting method. 18. The resolving power of a grating (i) depends on the size of the rulings. (ii) is independent of the spacing of the rulings. (iii) depends on the total number of rulings. (iv) depends on the diffraction order.

B. Problems 1. In the Fresnel diffraction geometry if you place a small circular obstacle (opaque) why do you observe a bright spot at the center (along the optic axis) of the shadow?

OUP CORRECTED – FINAL

Exercises 405 2. Using Fraunhofer diffraction, a grating with a large number of slits can be used to measure the wavelength of an unknown radiation accurately. Why? 3. Refraction: Show that when an incident wave (ray) passes through a medium, such as glass, of well-defined thickness and limited by plane parallel slides, the emergent wave is parallel to the incident ray. 4. Resolving power of a lens: A lens has a diameter of 4 cm and a focal length of 40 cm. It is illuminated by a beam of monochromatic light with λ = 560 nm. (a) What is the radius of the central disc observed in the diffraction pattern at the plane of focus? (b) What is the resolving power of this lens at this wavelength? 5. Resolving power of a grating: (a) Sodium has two yellow lines with wavelengths, λ1 = 589.0 nm and λ2 = 589.6 nm in the visible spectrum. To what transitions from the energy levels of sodium, discussed in §2, do these lines correspond? (b) Suppose you have a grating with 20,000 lines and a length of 0.04 m. Can this grating resolve the two yellow lines of sodium? 6. The Fraunhofer pattern for an ideal grating with N slits was given as I ≈ A20

sin2 β sin2 γ where β = (πw sin θ ) /λ and γ = (πd sin θ ) /λ β2 γ2 (6.9.19)

Make a qualitative sketch of the intensity pattern for FIVE equally spaced slits with d/w = 4. Label several points on the x-axis with corresponding values of β and γ . 7. Fresnel Diffraction: Consider a small hole, 1 mm in diameter, on an opaque screen illuminated by light of λ = 590.0 nm. We define the farthest point of darkness (FPD) as the point at which only two Fresnel zones are within the aperture. Now: (a) Calculate the distance along the optic axis from the screen to the FPD. (b) Would the FPD distance change with order? How? 8. A given point is vibrating with simple harmonic motion. Its period is 5.0 s and its amplitude is 0.03 m. If the initial phase angle is π/3 radians, find (a) the initial displacement. (b) the displacement after 12.0 s. (c) Plot this in a graph. 9. Ellipsometry: Light of wavelength, λ = 600 nm, is incident on a quartz crystal film, placed in air, at an incident angle of 45◦ . Calculate the reflection coefficient for s- and p-polarized light. What will these coefficients be for a Brewster angle incidence?

OUP CORRECTED – FINAL

406 Optics, Optical Methods, and Microscopy 10. Addition of simple harmonic motion at right angles in the y- and z-planes.Consider the two-component motion given by y = sin (ωt − α1 ) a1

(Ex6.10.1)

z = sin (ωt − α2 ) a2

(Ex6.10.2)

and

(a) (b) (c) (d) (e) (f) (g) (h) (i)

Expand equation (Ex6.10.1) to give (Ex6.10.3). Expand equation (Ex6.10.2) to give (Ex6.10.4). Multiply (Ex6.10.3) by sin α 2 to give (Ex6.10.5). Multiply (Ex6.10.4) by sin α 1 to give (Ex6.10.6). Subtract (Ex6.10.5) from (Ex6.10.6) to give (Ex6.10.7). Similarly, multiply (Ex6.10.3) by cos α 2 to give (Ex6.10.8). Multiply (Ex6.10.4) by cos α 1 to give (Ex6.10.9). Subtract (Ex6.10.9) from (Ex6.10.8) to give (Ex6.10.10). Now, square both (Ex6.10.7) and (Ex6.10.10) to give (Ex6.10.11) and (Ex6.10.12). (j) Add (Ex6.10.11) and (Ex6.10.12) and show that this gives the solution sin2 (α1 − α2 ) =

z2 2yz y2 + − cos(α1 − α2 ) a1 a2 a12 a22

(Ex6.10.2)

11. Compare the resolution and depth of field of an optical and a transmission electron microscope. Assume that we are using blue light for the optical microscope. For the electron microscope assume an acceleration voltage of 200 kV and the angle α = 5◦ . Make any other relevant assumptions and state them clearly. 12. Consider the following two waves propagating on orthogonal planes Ex = ˆia1 sin(kz − ωt) and Ey = ˆja2 sin(kz − ωt + δ) where δ is the relative phase difference between them. (a) If δ = 0 or mπ, where m is an integer, show that the superposition of these two waves gives rise to linear polarization for both even and odd values of m. (b) If δ = −π/2 + 2mπ, where m = 0, ±1, ±2, . . . ., and a = a1 = a2 , what is the resultant wave? What happens to the resultant wave, at some point z0 , as a function of time? Hint: Use t = 0, and t = kz0 /ω to discuss your answer. (c) If δ = π/2 + 2mπ, where m = 0, ±1, ±2, . . . ., and a = a1 = a2 , what is the resultant wave at some point z0 , as a function of time? What is the difference between (b) and (c)?

OUP CORRECTED – FINAL

Exercises 407 13. Read the paper by Paterson et al. [4] and compare laser scanning confocal microscopy with scanning electron microscopy (§10), especially in the context of characterizing the morphology of hydrogels. 14. A thin film of copper, with refractive index, nCu = 0.95, placed in air (nair = 1.0), is investigated with a polarized laser of wavelength, λ = 550 nm. For an incident angle, θi = 45◦ , in air, calculate: (a) the transmitted angle, θ t , in Cu. (b) the amplitude reflection coefficient for p-polarized (rp ) and s-polarized (rs ) light. (c) the amplitude ratio (ρ). (d) the difference r = |rs— rp |. (e) Now calculate and plot the variation of r as a function of θ i for 0 < θ i < 90◦ . (f) The Brewster angle is defined as the value of θ I when r is a maximum. Determine the Brewster angle from your plot. (g) Compare with the estimate of the Brewster angle given by (6.9.14) and comment on the result.

OUP CORRECTED – FINAL

X-Ray Diffraction

7 7.1 Introduction

409

7.2 Interaction of X-Rays with Electrons

409

(a)

(b)

Base pairs

7.3 Scattering by an Atom: Atomic Scattering Factor

415

Adenine Thymine

7.4 Scattering by a Crystal: Structure Factor

422

Guanine Cytosine

7.5 Examples of Structure Factor Calculations

425

7.6 Symmetry and Structure Factor

431

7.7 The Inverse Problem of Determining Structure from Diffraction Intensities

432

Sugar phosphate backbone

7.10 Factors Influencing X-Ray Diffraction Intensities

451

7.11 Applications of X-Ray Diffraction

459

The “double helix” structure (a) of DNA, first proposed by Crick and Watson [1], is now universally accepted. However, this X-ray diffraction photograph (b) by Franklin and Gosling [2] of a DNA fiber containing many millions of aligned DNA strands was absolutely crucial in establishing the correctness of the structural model proposed by Crick and Watson. The two arms of spots are characteristic of the helical structure, and the angle between them represents the ratio of the width of the molecule to the repeat distance of the helix. The fourth spot along each arm is missing, which indicates that the two helices are intertwined.

Summary

471

(Figure credit: US National Library of Medicine).

Further Reading

472

References

473

Exercises

474

7.8 Broadening of Diffracted Beams and Reciprocal Lattice Points

433

7.9 Methods of X-Ray Diffraction

437

Principles of Materials Characterization and Metrology. Kannan M. Krishnan, Oxford University Press (2021). © Kannan M. Krishnan. DOI: 10.1093/oso/9780198830252.003.0007

OUP CORRECTED – FINAL

Interaction of X-Rays with Electrons 409

7.1 Introduction In this chapter, we turn our attention to one of the important and widely used methods of materials characterization: X-ray diffraction. Before one can tackle this subject in some detail, it is important to have a good understanding of the elementary principles of diffraction and crystallography introduced in §4. This includes concepts of the reciprocal lattice (§4.2), Bragg’s law and the Laue condition (§4.3) for diffraction, as well as optics, especially the description of simple harmonic waves (§6.2) and optical diffraction (§6.6). All these topics have been presented earlier. Here, we make our way systematically by first describing the interaction of X-rays with electrons, then their scattering by atoms to define the atomic scattering factor, followed by their scattering by the unit cell to develop the concept of structure factors, and finally, their relationship to the observed diffraction intensities. It then describes various methods of X-ray diffraction popular in laboratory settings for both powder and single crystal specimens. The chapter describes the main factors influencing X-ray diffraction intensities, including temperature, absorption, multiplicity, and Lorentz polarization, and then introduces typical applications of X-ray diffraction in materials characterization: accurate measurement of lattice parameters, grain size, and lattice strain measurements; identification of crystallographic phases and the refinement of their structures; and order-disorder phase transitions. At the end, the chapter illustrates the versatility of X-ray diffraction with brief descriptions of in situ measurements in a synchrotron, and a miniaturized instrument sent to Mars by NASA to analyze the structure of minerals on the Martian surface. It is important to emphasize that X-ray diffraction finds widespread use not only in materials sciences and engineering (MSE), but also in physics, chemistry, biology, geology, art conservation, etc., and as such it is an invaluable method. Further, a good understanding of X-ray diffraction will make it easy to appreciate related methods of electron and neutron diffraction, which are discussed in Chapter 8. There are far too many books on X-ray diffraction, each treating the subject from a different perspective. In writing this chapter, I have particularly benefitted from Cullity (1978), de Graef and McHenry (2007), Giacovazzo et al. (2007), Hammond (2006), Klug and Alexander (1974), Schwartz and Cohen (1987), Warren (1990), and Woolfson (1997). Many more details that cannot be included in a single chapter are to be found in these comprehensive texts on X-ray diffraction, and the reader is encouraged to consult them as appropriate.

7.2 Interaction of X-Rays with Electrons When an X-ray beam interacts with an atom two processes are possible. X-ray photons may be absorbed with either the emission of a photoelectron (§2.3) forming the basis of X-ray photoemission spectroscopy (§3.8), or following an inner-shell ionization the emission of a characteristic X-ray photon (§2.5.2.1)

OUP CORRECTED – FINAL

410

X-Ray Diffraction forming the basis of X-ray fluorescence spectroscopy (XRF). Alternatively, the X-ray beam may be scattered (§1.4.2), and such scattering may also include two different components. First, we observe an unmodified scattered wave with the same wavelength as the primary beam of incident X-rays. This is referred to as Thomson1 scattering and can be explained classically as follows. The primary X-ray beam is an electromagnetic wave (§6.2) with the electric vector, E, at any point varying sinusoidally with time, and directed in a plane orthogonal to the direction of propagation. When this wave encounters an electron—a charged particle—the time-varying electric field of the incident wave will exert a force on the electron and set the latter also in oscillation around its mean position. This causes the electron to be accelerated and decelerated, and in classical electromagnetism such a repetitively accelerated charge will emit an electromagnetic wave. We refer to this absorption of the incident wave and its re-emission at the same wavelength as scattering. Further, we observe that this scattering of X-rays by electrons is coherent, by which we mean that there is a definite relationship between the phase of the scattered and incident X-ray waves. In the case of scattering of X-rays by electrons, this phase shift is λ/2; since all the electrons of the atom scatter with the same phase-shift, the scattered X-ray waves are coherent. Finally, even though the X-rays are scattered by an electron in all directions, the intensity of the scattered wave depends only on the angle of scattering. Second, we also observe a modified scattered wave, with a wavelength longer than that of the incident X-rays. This Compton2 scattering is incoherent with the incident wave (i.e. the phase relationship between them is random), and can only be explained in a quantum mechanical framework, including the principle of wave-particle duality introduced in §1.3.3. Now, the incident X-ray beam is considered to be a stream of X-ray quanta (photons), each of energy, E1 = h f1 , and their interactions with the electron is best described in terms of billiard balllike elastic collisions with a conservation of both energy and momentum. In this scenario, it can be shown that the change of wavelength of the scattered X-rays is independent of the wavelength of the incident radiation, but is dependent only on the scattering angle. We now describe both the classical Thomson (coherent) and Compton (incoherent) scattering in detail, and discuss their relevance to X-ray diffraction from materials.

1 Sir J. J. Thomson (1856–1940) was an English physicist who received the 1906 Nobel prize in Physics and cited for “his theoretical and experimental investigations on the conduction of electricity by gases”. 2 A. H. Compton (1892–1962) was a US physicist who received the 1927 Nobel prize in Physics and cited for “discovering the effect named after him”.

7.2.1 Thomson Coherent Scattering We consider an electron at the origin, O, interacting with an unpolarized primary X-ray beam of intensity, I 0 , traveling along the x-axis (Fig. 7.2.1). Our goal is to determine the intensity, I2θ , of the scattered wave at the position, P, at a distance, R, from the electron, and at an angle, 2θ , from the x-axis. Note that the point, P, is in the xy-plane. Further, since the incident beam is unpolarized, its electric field, E,

OUP CORRECTED – FINAL

Interaction of X-Rays with Electrons 411 z E0z

E0

O

E0y 2θ

I0 y E

R

Ey

Figure 7.2.1 Classical scattering of an incident electromagnetic radiation of intensity, I 0 , by an electron at O. The components of the electric vector of the scattered and incident radiations are shown.

Ez P

x can take all possible orientations in the yz-plane with equal probability. Without loss of generality, we first consider any one orientation, E0 , with components E 0y and E 0z , as shown, and later average it over all directions. We also know from the theory of electromagnetism that a charge, e.g. an electron, at position, O, subject to an acceleration, a (Fig. 7.2.2) will generate an electromagnetic radiation at any point, P, with an electric vector, E, of amplitude ea E= sin α 4π ε0 Rc2

(7.2.1)

az =

E0z e m

(7.2.2a)

ay =

E0y e m

(7.2.2b)

and

As a result, at any point, P, applying (7.2.1), we find the electric vector components of the scattered wave as e2 E0z 4π ε0 Rc2 m

(7.2.3a)

e2 E0y cos 2θ 4π ε0 Rc2 m

(7.2.3b)

and Ey =

a α

R

P

e– O

where c is the velocity of light. The direction of E is always perpendicular to OP and lies in the plane of OP and a. When an X-ray beam is incident on the electron (Fig. 7.2.1), its electric field component will set the electron in oscillation, with acceleration components

Ez =

E

Figure 7.2.2 The relationship of a scattered electromagnetic wave at position, P, arising from the acceleration of an electron at O. Both vectors, a and E, are in the same plane. Simply put, the component, a sin α, of the acceleration seen by an eye placed at the point, P, of observation will determine the electric field produced.

OUP CORRECTED – FINAL

412

X-Ray Diffraction Then, the net amplitude, E, of the scattered radiation at the point, P, is given by  E 2 = Ez2 + Ey2 =

e2 4π ε0 Rc2 m

2 

2 2 + E0y cos2 2θ E0z

 (7.2.4)

Now, since the initial beam is unpolarized, we take directional averages, i.e. 

     2 2 E02 = E0y + E0z

(7.2.5)

Further, since the y- and z-axis are equivalent, we get 

   1  2 2 E2 E0y = E0z = 2 0

(7.2.6)

Thus, (7.2.4) can be modified as    E 2 = E02



e2 4π ε0 Rc2 m

2 

1 + cos2 2θ 2

(7.2.7)

Since the observable quantity is the intensity, I2θ , of the scattered radiation we can

 set E02 = I0 and rewrite (7.2.7) to obtain the intensity of the classical scattering of an electromagnetic wave by a free electron as  I2θ = I0

e2 4πε0 Rc2 m

2 

 κ 2 1 + cos2 2θ 1 + cos2 2θ = I0 2 R 2

 This is the classical Thomson scattering equation. The quantity, κ =

(7.2.8) e2 4πε0 c2 m

 ,

with the dimension of length, equals 2.82 × m, and in classical electromagnetic theory is referred to as the radius of the electron. Note that the scattered intensity ratio, I2θ /I0 , at a distance of a few centimeters from the electron, is of the order of 10–26 . Even though this intensity ratio is very small, in practice the number of electrons even in a very small quantity (mg) of the specimen is ∼1020 , and hence we get a measurable signal when all their contributions are combined. Further, even the lightest nucleus (a proton) has a mass ∼1,840 times that of the electron. Since the scattered intensity is inversely proportional to the square of the mass, we expect only the electrons to effectively scatter the X-rays. In other words, even though the nucleus has a charge, because of its very high mass, it contributes  negligibly to X-ray scattering. The factor 1 + cos2 2θ /2 in (7.2.8) is referred to as the polarization factor; this nomenclature is ironic as it arises because of our assumption of an unpolarized incident beam. Finally, as mentioned earlier, such Thomson scattering by electrons is coherent, with a uniform phase different of π between the scattered and incident radiation. 10–15

OUP CORRECTED – FINAL

Interaction of X-Rays with Electrons 413 We are ultimately interested in the scattering of X-rays by atoms; in §7.3, we describe how the coherent scattering of the individual electrons are combined to describe the intensity of scattering by the atom, and later in §7.4, by the periodic arrangements of atoms in a crystalline unit cell. Example 7.2.1: How can Thomson scattering be used to polarize and analyze X-rays? Solution: We use the simple rule illustrated in Figure 7.2.2 and consider a beam of unpolarized X-rays incident along the X-axis on a scatterer (block of Carbon) at position, O, in the figure below. Z Y O

X Y

O

X

The electrons in the scatterer are accelerated in all directions in the Y–Z plane. However, if we place an eye at the point O , we will only see the y-component of the electron acceleration. Thus, any X-rays scattered in the direction of O–O , i.e. through a scattering angle of 90◦ , will be linearly polarized (i.e. produces a polarized beam). Now, if we place another scatterer at O , all its electrons will be accelerated only along the Y direction. If we place the eye at Y we will see no component of the Y -acceleration, but at X we will see a maximum. In summary, the first scattering of 90◦ at O produces a linearly polarized beam, and the second scattering at 90◦ at O serves as the analyzer. Adapted from Warren (1990).

7.2.2 Compton Incoherent Scattering To describe Compton scattering, it is best to invoke the quantum mechanical principle of wave–particle duality (§1.3.3), and describe the elastic3 interaction as a collision of an X-ray photon and an electron. Figure 7.2.3a shows the geometry of the scattering, with the incident photon traveling along AO, interacting with the electron at O, and after the collision moving along OP with a small change in wavelength,  λ; the electron recoils and moves along OE.

3 In some texts, Compton scattering is called an inelastic collision. Here, they are referring only to the fact that the energy of the incident and scattered photons is different, and ignore the kinetic energy of the recoil electron.

OUP CORRECTED – FINAL

414

X-Ray Diffraction (a)

Figure 7.2.3 (a) The geometry of scattering of the X-ray photon and an electron. (b) The momentum vector diagram involved in Compton scattering.

A

Scattered Photon, λ+λ Incident Photon, λ

O

P

(b) h λ+λ

Recoil Electron, me, ve

θ h λ

meve

E

For an elastic collision, the total energy is conserved. This gives hc 1 hc = + me v2e λ λ + λ 2

(7.2.9)

1 hc λ = me v2e 2 λ2

(7.2.10)

or, by rearranging terms

Further, the momentum (Fig. 7.2.3b) is also conserved in the elastic collision. This gives h 1 sin θ = me ve λ 2

(7.2.11)

Eliminating ve between the two equations (7.2.10) and (7.2.11) gives λ =

2h h ˚ sin2 θ = (1 − cos 2θ ) = 0.024 (1 − cos 2θ ) A me c me c

(7.2.12)

where the physical constants have been replaced with their actual values. Note that the change in the wavelength, λ, associated with Compton scattering is independent of the initial wavelength, and the maximum value of λ is expected for 2θ = π , with a value, λ = 0.048, which is quite significant for the scattering of X-ray photons with λ ∼ 1 Å. However, unlike Thomson scattering, the phase of the Compton-scattered wave has no relationship to the incident wave. As a result, it contributes insignificantly to the diffraction intensity; but it does contribute some background intensity in diffraction patterns. In general, both Thomson and Compton scattering take place, and one must resort to a complete Quantum Mechanical theory to accurately account for the coherent and incoherent scattering contributions. However, even in such a full Quantum Mechanical treatment, the total intensity of the scattered wave, per electron, taking both Thomson and Compton contributions into account, is closely equal to the value given by the classical Thomson equation (7.2.8).

OUP CORRECTED – FINAL

Scattering by an Atom: Atomic Scattering Factor 415

7.3 Scattering by an Atom: Atomic Scattering Factor Let us first consider the classical scattering of an X-ray beam by two electrons confined to a small volume in atomic orbit (Fig. 7.3.1). In the forward direction (Fig. 7.3.1a), 2θ = 0, both electrons will scatter the incident X-ray wave by an identical phase difference of λ/2. More importantly, at distances sufficiently far from the two electrons the path length of the wave scattered by the two electrons is the same; they arrive in phase and hence their amplitudes can be added directly to give the scattered wave amplitude. By extension, for an atom with Z electrons (only two are shown in the figure), the forward scattered wave amplitude is Z times the amplitude of the wave scattered by a single electron. In any other direction, scattering of the incident X-ray by electrons in different positions introduce changes in the phase of the scattered wave because of their path length difference (such as A BB , for the two electrons shown in Figure 7.3.1b). The difference, typically less than a wavelength, results in partial coherence of the waves in the scattered direction, at any distance, R, far from the two electrons, causing a decrease in amplitude when compared to the scattered wave in the forward direction. We can represent the scattering conditions in vector form (Fig. 7.3.1c). For the point, P, at a distance, R, far from the electrons, the primary and scattered wave directions can be represented by the unit vectors, s0 and s, respectively. Then, the total path length, l 1 + l 2 , for the scattered plane wave from any electron at rn , observed at the position, P, is s0 + R − rn · s = R − ( s − s0 ) · rn l1 + l2 = rn ·

(7.3.1)

Following Thomson scattering, the scattered wave in complex notation (see §6.2.4), with frequency, f, can be written as P (a)

(b)

(c) R

A

A

B

A B

R

2 B

s0 s O rn

l2

l1

Figure 7.3.1 Scattering of an incident X-ray beam by two representative electrons in an atom in (a) forward scattering direction, and (b) a direction at an angle, 2θ, with a path difference, A BB , between the waves scattered by electrons at positions A and B. (c) The scattering geometry in vector form, showing the path length, l 1 + l 2 , for the wave scattered by the electron at position, rn , and observed at position, P, at a distance, R, far from O.

OUP CORRECTED – FINAL

416

X-Ray Diffraction E = E0

    κ κ R − ( s − s0 ) · rn (l1 + l2 ) exp 2π i ft − = E0 exp 2πi ft − R λ R λ (7.3.2) 2

where κ = 4πεe c2 m , as defined in (7.2.8). At position, P, the resultant electric field 0 component, EP , of the scattered wave, for n electrons can be written as the sum of the waves scattered from the individual electrons as

 κ 2π i R  [( (7.3.3) exp EP = E0 exp 2π i ft − s − s0 ) · rn ] R λ n λ Now, instead of assuming that each electron is a localized scatterer of X-rays, we consider that each electron is spread out as a diffuse cloud of charge with a charge density, ρ, expressed in electron units per unit volume. Then, ρ dV is the ratio of the charge in any volume unit, dV, normalized to the charge of one electron.  Thus, for a single electron, ρ dV = 1. To get the contribution for the scattering from a single electron, we must integrate over the volume, dV, making sure that we account for the phase difference for scattering from each charge element dV. Then the sum in (7.3.3) is replaced by the integral EP = E0

   κ R 2π i [( exp 2πi ft − exp s − s0 ) · r] ρdV R λ λ

(7.3.4)

where the individual electrons at positions, rn , are replaced by charge elements ρdV at r. The quantity represented by the integral  fe =

s - s0 φ

r

s

2θ s0 O Figure 7.3.2 For the atom centered at O, the relationship between ( s − s0 ) and r.

exp

2π i [( s − s0 ) · r] ρdV λ

(7.3.5)

is called the electron scattering factor, and represents the ratio of scattering by an electron with distributed charge normalized by the scattering from a single electron in the classical theory. As a first approximation, we will assume that the charge distribution for each electron in the atom is spherically symmetric, i.e. ρ = ρ(r). Then, referring to the geometry in Figure 7.3.2, we can show that ( s − s0 ) · r = 2 sin θ r cos φ, and dV = 2πr 2 sin φdφdr. We can now rewrite (7.3.5) as ∞ π fe = r=0φ=0

 4π sin θ exp i r cos φ ρ(r)2π r 2 sin φdφdr λ

(7.3.6)

We now set s = sinλ θ , and carry out the integral over φ. Thus ∞ fe =

r2 r=0

sin 4π sr ρ(r)dr sr

(7.3.7)

OUP CORRECTED – FINAL

Scattering by an Atom: Atomic Scattering Factor 417 Finally, for an atom with n electrons, we add their individual contributions to get

fatom =



fe n =

n

∞  n

r2

sin 4π sr ρn (r)dr sr

(7.3.8)

r=0

The simple number, fatom , is called the atomic scattering factor and represents the ratio fatom =

ampitude of the wave scattered by the atom amplitude of the wave scattered by a single electron

(7.3.9)

It can be computed if the radial distribution of the electron density in the atom,  n ρn (r) is known. Note that fatom is a function of the parameter s, and hence it varies as a function of sinλ θ . Figure 7.3.3a shows typical plots of fAu and fCu . It is clear from the plot that for both elements, fatom (0) = Z, where the number of electrons is equal to the atomic number, Z. In other words, for each element the dimensionless atomic scattering factor begins at Z for forward scattering (θ = 0) and then decreases with scattering angle to a very low value for backward scattering (2θ = π ); it also decreases with decreasing wavelength, λ, of the incident X-rays. Figure 7.3.3b shows the atomic scattering factor for O2– , Ne, Si4+ , all three containing 10 electrons (Z = 10). Again, for all three ions, f (0) = Z = 10, but then the scattering factor falls off below 10 for larger values of scattering angles. However, the rate at which they fall with angle is dependent on the relative size of the atom or ion. Of the three, O2– is the largest in size and hence the phase difference for scattering between the electrons is largest for large scattering angles, leading to the most destructive interference. At the other end, Si4+ is smallest in size, with Ne in between, and this is also reflected in the angular dependence. Tables for atomic scattering factors for all the elements are readily available in two forms. In many textbooks, they are listed in tabular form as a function of sin θ λ , typically in steps of 0.1, from 0 to 1.2. We do the same here in Table 7.3.1. Note that the tabulated values of fatom are strictly valid only when the wavelength of the scattered radiation is much smaller than the absorption edge of the scattering atom; if the two values are close, a small correction called the anomalous dispersion correction, must be applied (see Warren, 1990). Alternatively, the atomic scattering factors are parametrized, as a function of sinλ θ (= s) in terms of curve fitting parameters, using four exponential functions. In this case, the scattering for any value of s can be computed from fatom (s) =

4  i=1

  ai exp −bi s2 + c

(7.3.10)

OUP CORRECTED – FINAL

418

X-Ray Diffraction 80 10

(a)

(b)

Au 8

60

Si4+

fatom

fatom

6 40

Ne 4

O2–

Cu 20 2

0 0.0

0.2

0.4 0.6 sin θ (Å–1) λ

0.8

1.0

0 0

0.2

0.4 0.6 sin θ (Å–1) λ

0.8

Figure 7.3.3 The atomic scattering factor as a function of sinλ θ for (a) two elements, Au and Cu, and (b) for three isoelectronic atoms and ions with 10 electrons. ˚ (= 0.1 nm), s is determined where the wavelength is expressed in units of A from the diffraction angle, θ hkl , and the corresponding terms are tabulated, as the Cromer–Mann parameters [3], for some representative elements in Table 7.3.2. We have determined the coherently scattered wave amplitude by considering that the electronic charge distribution in an atom is distributed in space and not localized at a specific point. In quantum mechanics, the wave function, , describes the state of the electron and is related to the electron distribution, ρ, simply as ρ = | |2 . In the case of a spherically symmetric potential, such as that experienced by an electron in an atom, the wave function can be shown to be a product of a radial component, R(r), and an angular component, Y(θ, φ), determining the shape of the orbital, i.e. = R(r) Y (θ , φ). Thus, our assumption of spherical symmetry included in the electron distribution, ρ(r), is not unreasonable, especially for a completely filled shell. Further, electrons occupy discrete energy levels in atoms. Thus, Thomson scattering must result in no change in the energy level of the electron; in other words, the electron must be so tightly bound that no momentum is transferred to it upon impact. In contrast, Compton scattering must involve either well-defined atomic transitions between bound states or the complete ejection of the electron from the atom. In summary, the complete quantum mechanical treatment shows that the intensities of the incoherent scattering become significant only for large values

Table 7.3.1 X-ray Atomic Scattering Factors (dimensionless numbers). Adapted from Warren (1990) 0.0

0.1

0.2

0.3

0.4

0.5

0.6

H He Li Be B C N O F Ne Na Na+ Mg Al Si P S Cl Cl– A K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se

1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 10.0 12.0 13.0 14.0 15.0 16.0 17.0 18.0 18.0 19.0 20.0 21.0 22.0 23.0 24.0 25.0 26.0 27.0 28.0 29.0 30.0 31.0 32.0 33.0 34.0

0.81 1.83 2.22 3.07 4.07 5.13 6.20 7.25 8.29 9.36 9.76 9.55 10.50 11.23 12.16 13.17 14.33 15.33 16.02 16.30 16.73 17.33 18.72 19.41 20.47 21.93 22.61 23.68 24.74 25.80 27.19 27.92 28.65 29.52 30.47 31.43

0.48 1.45 1.74 2.07 2.71 3.58 4.60 5.63 6.69 7.82 8.34 8.39 8.75 9.16 9.67 10.34 11.21 12.00 12.20 12.93 13.73 14.32 15.39 16.07 17.03 18.37 19.06 20.09 21.13 22.19 24.63 24.33 24.92 25.53 26.20 26.91

0.25 1.06 1.51 1.71 1.99 2.50 3.24 4.09 5.04 6.09 6.89 6.93 7.46 7.88 8.22 8.59 8.99 9.44 9.40 10.20 10.97 11.71 12.39 13.20 14.03 15.01 15.84 16.77 17.74 18.73 19.90 20.77 21.47 22.11 22.69 23.24

0.13 0.74 1.27 1.53 1.69 1.95 2.40 3.01 3.76 4.62 5.47 5.51 6.20 6.77 7.20 7.54 7.83 8.07 8.03 8.54 9.05 9.64 10.12 10.83 11.51 12.22 13.02 13.84 14.68 15.56 16.48 17.42 18.26 19.02 19.69 20.28

0.07 0.52 1.03 1.37 1.53 1.69 1.94 2.34 2.88 3.54 4.29 4.33 5.01 5.69 6.24 6.67 7.05 7.29 7.28 7.56 7.87 8.26 8.60 9.12 9.63 10.14 10.80 11.47 12.17 12.91 13.65 14.51 15.38 16.19 16.95 17.63

0.04 0.36 0.82 1.20 1.41 1.54 1.70 1.94 2.31 2.79 3.40 3.42 4.06 4.71 5.31 5.83 6.31 6.64 6.64 6.86 7.11 7.38 7.64 7.98 8.34 8.72 9.20 9.71 10.26 10.85 11.44 12.16 12.95 13.72 14.48 15.20

1 2 3 4 5 6 7 8 9 10 11 11 12 13 14 15 16 17 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

0.7

0.8

0.02 0.02 0.25 0.18 0.65 0.51 1.03 0.88 1.28 1.15 1.43 1.32 1.55 1.44 1.71 1.57 1.96 1.74 2.30 1.98 2.76 2.31 2.77 2.31 3.30 2.72 3.88 3.21 4.47 3.75 5.02 4.28 5.56 4.82 5.96 5.27 5.97 5.27 6.23 5.61 6.51 5.95 6.75 6.21 6.98 6.45 7.22 6.65 7.48 6.86 7.75 7.09 8.09 7.32 8.47 7.60 8.88 7.91 9.33 8.25 9.80 8.61 10.37 9.04 11.02 9.54 11.68 10.08 12.37 10.67 13.06 11.27

0.9

1.0

1.1

1.2

1.3

1.4

1.5

0.01 0.13 0.40 0.74 1.02 1.22 1.35 1.46 1.59 1.76 2.00 2.00 2.30 2.71 3.16 3.64 4.15 4.60 4.61 5.01 5.39 5.70 5.96 6.19 6.39 6.58 6.77 6.99 7.22 7.48 7.76 8.08 8.46 8.87 9.34 9.83

0.01 0.10 0.32 0.62 0.90 1.11 1.26 1.37 1.48 1.61 1.78 1.79 2.01 2.32 2.69 3.11 3.56 4.00 4.00 4.43 4.84 5.19 5.48 5.72 5.94 6.14 6.32 6.51 6.70 6.90 7.13 7.37 7.64 7.96 8.32 8.71

0.07 0.26 0.52 0.78 1.01 1.18 1.30 1.40 1.50 1.63 1.63 1.81 2.05 2.35 2.69 3.07 3.47 3.47 3.90 4.32 4.69 5.00 5.29 5.53 5.74 5.93 6.12 6.29 6.47 6.65 6.84 7.05 7.29 7.57 7.86

0.05 0.21 0.43 0.68 0.91 1.08 1.22 1.32 1.42 1.52 1.52 1.65 1.83 2.07 2.35 2.66 3.02 3.03 3.43 3.83 4.21 4.53 4.84 5.10 5.34 5.54 5.74 5.91 6.08 6.25 6.42 6.58 6.77 6.98 7.21

0.04 0.16 0.37 0.60 0.82 1.01 1.14 1.25 1.35 1.44 1.44 1.54 1.69 1.87 2.10 2.34 2.65 2.65 3.03 3.40 3.77 4.09 4.41 4.71 4.94 5.18 5.39 5.58 5.75 5.90 6.07 6.21 6.37 6.54 6.72

0.03 0.03

0.74 0.66

1.28 1.22 1.37 1.31 1.37 1.30 1.57 1.48 1.71 1.60 1.89 1.75

2.35 2.11 3.01 3.37 3.68 4.01 4.30 4.55 4.80 5.03 5.23 5.41 5.57 5.73 5.88 6.02 6.17 6.31

2.71 3.03 3.31 3.64 3.93 4.18 4.45 4.69 4.90 5.09 5.25 5.43 5.58 5.72 5.86 5.99

continued

OUP CORRECTED – FINAL

(sin θ )/λ(Å−1 )

(sin θ )/λ(Å−1 )

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

1.3

1.4

Br Kr Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I Xe Cs Ba Ta W Re Os Ir Pt Au Hg Tl Pb Bi Th U

35.0 36.0 37.0 38.0 39.0 40.0 41.0 42.0 43.0 44.0 45.0 46.0 47.0 48.0 49.0 50.0 51.0 52.0 53.0 54.0 55.0 56.0 73.0 74.0 75.0 76.0 77.0 78.0 79.0 80.0 81.0 82.0 83.0 90.0 92.0

32.43 33.44 34.11 35.06 36.01 36.96 37.91 38.86 39.81 40.76 41.72 42.67 43.63 44.58 45.5 46.5 47.5 48.4 49.4 50.3 51.3 52.3 68.6 69.5 70.5 71.5 72.4 73.4 74.4 75.3 76.3 77.2 78.2 85.0 86.9

27.70 28.53 28.97 29.83 30.68 31.54 32.40 33.25 34.12 34.98 35.84 36.70 37.57 38.44 39.3 40.2 41.1 41.9 42.8 43.7 44.5 45.4 60.4 61.3 62.2 63.1 64.0 64.9 65.8 66.7 67.6 68.5 69.3 75.6 77.4

23.82 24.40 24.75 25.51 26.28 27.04 27.81 28.57 29.34 30.12 30.89 31.67 32.44 33.22 34.0 34.8 35.6 36.4 37.1 37.9 38.7 39.5 53.1 54.0 54.8 55.6 56.4 57.2 58.0 58.8 59.7 60.5 61.3 67.1 68.7

20.84 21.34 21.29 21.96 22.64 23.32 24.01 24.69 25.38 26.07 26.76 27.46 28.16 28.85 29.6 30.3 31.0 31.7 32.4 33.1 33.8 34.5 46.9 47.6 48.3 49.1 49.8 50.6 51.3 52.1 52.8 53.6 54.3 59.6 61.1

18.27 18.82 18.55 19.15 19.76 20.37 20.98 21.60 22.21 22.83 23.46 24.08 24.71 25.34 26.0 26.6 27.2 27.9 28.5 29.2 29.8 30.4 41.7 42.3 43.0 43.7 44.4 45.0 45.7 46.4 47.1 47.8 48.5 53.3 54.7

15.91 16.54 16.30 16.84 17.39 17.94 18.49 19.04 19.60 20.16 20.72 21.28 21.85 22.42 23.0 23.6 24.1 24.7 25.3 25.9 26.5 27.0 37.3 37.9 38.5 39.1 39.7 40.3 41.0 41.6 42.2 42.9 43.5 47.9 49.2

13.78 14.44 14.47 14.96 15.46 15.95 16.43 16.95 17.46 17.96 18.47 18.98 19.50 20.02 20.5 21.1 21.6 22.1 22.6 23.2 23.7 24.2 33.6 34.1 34.7 35.3 35.8 36.4 37.0 37.5 38.1 38.7 39.4 43.3 44.5

11.93 12.57 12.94 13.39 13.84 14.29 14.74 15.20 15.65 16.12 16.58 17.05 17.52 17.99 18.5 18.9 19.4 19.9 20.4 20.9 21.3 21.8 30.4 30.9 31.4 32.0 32.5 33.0 33.5 34.1 34.6 35.1 36.1 39.4 40.5

10.41 10.97 11.66 12.07 12.48 12.89 13.31 13.73 14.15 14.57 14.99 15.42 15.85 16.28 16.7 17.2 17.6 18.0 18.5 18.9 19.4 19.8 27.7 28.2 28.7 29.1 29.6 30.1 30.6 31.1 31.6 32.1 33.1 36.1 37.1

9.19 9.66 10.58 10.95 11.32 11.70 12.08 12.46 12.85 13.24 13.63 14.02 14.42 14.81 15.2 15.6 16.0 16.4 16.8 17.2 17.7 18.1 25.4 25.8 26.3 26.7 27.1 27.6 28.0 28.5 29.0 29.4 30.5 33.1 34.0

8.24 8.62 9.65 9.99 10.34 10.68 11.04 11.39 11.74 12.10 12.46 12.82 13.19 13.56 13.9 14.3 14.7 15.1 15.4 15.8 16.2 16.6 23.3 23.7 24.2 24.6 25.0 25.4 26.3 26.3 26.7 27.1 25.5 30.5 31.4

7.51 7.81 8.84 9.16 9.48 9.80 10.13 10.45 10.78 11.11 11.45 11.78 12.12 12.46 12.8 13.2 13.5 13.8 14.2 14.5 14.9 15.3 21.6 21.9 22.3 22.7 23.1 23.5 24.3 24.3 24.7 25.1 23.6 28.3 29.1

6.95 7.19 8.14 8.44 8.73 9.03 9.33 9.64 9.94 10.25 10.56 10.87 11.19 11.51 11.8 12.1 12.5 12.8 13.1 13.4 13.8 14.1 20.0 20.3 20.3 21.1 21.4 21.8 22.5 22.5 22.9 23.3 22.0 26.3 27.0

6.51 6.70 7.53 7.80 8.08 8.36 8.64 8.92 9.21 9.49 9.78 10.07 10.37 10.66 11.0 11.3 11.6 11.9 12.2 12.5 12.8 13.1 18.6 18.9 18.9 19.6 19.9 20.3 21.0 21.0 21.3 21.7 22.0 24.5 25.2

35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 73 74 75 76 77 78 79 80 81 82 83 90 92

1.5 6.16 6.31 6.99 7.24 7.50 7.76 8.02 8.29 8.55 8.82 9.09 9.37 9.64 9.92 10.2 10.3 10.8 11.0 11.3 11.6 11.9 12.2 17.4 17.7 17.7 18.3 18.6 18.9 19.6 19.6 19.9 20.3 20.6 22.9 23.6

OUP CORRECTED – FINAL

Table 7.3.1 Continued

OUP CORRECTED – FINAL

Scattering by an Atom: Atomic Scattering Factor 421 Table 7.3.2 Cromer–Mann atomic scattering parameters for select elements and ions [3]. a1

Element Atomic # C O2– Na+ Si Si4+ Cl– Cr Fe Cu Cs+ Au

6 8 11 14 14 17 24 26 29 55 79

a2

a3

a4

2.3 1.02 1.589 4.758 3.837 0 3.256 3.936 1.3998 5.7941 3.224 2.428 4.4392 3.2035 1.1945 18.292 7.2084 6.5337 10.6406 7.3537 3.324 11.9185 7.0485 3.3433 13.338 7.1676 5.6158 23.965 21.2204 9.7673 37.3027 14.9306 10.3425

b1

b2

0.865 20.844 10.208 0 7.831 30.05 1.0032 2.6671 6.1153 1.3215 2.5761 34.1775 0.4165 1.6417 3.4376 2.3386 0.0066 1.1717 1.4922 6.1038 0.392 2.2723 4.8739 0.3402 1.6735 3.5828 0.247 1.6155 0.2045 3.4388 2.0123 1.0081 6.5255

Adapted from Graef and McHenry (2007).

of s (> 1). However, for values of s (< 1) relevant for X-ray diffraction, the classical Thomson coherent scattering is a very good approximation of the observed intensities. Finally, in a crystal, the amplitudes of the coherent scattering from different atoms are added together; on the other hand, if the scattering is incoherent, it is only the intensities that add. As a result, we ignore the incoherent scattering and consider the coherent scattering as a very good approximation of the observed intensities. Example 7.3.1: Calculate the scattering factor for a sphere of uniform charge density, ρ0 , of radius, R0 . How will the intensity of scattering vary as a function of R0 ? Can you use this variation to determine the radius R0 ? Solution: (a) The scattering factor for a sphere of uniform charge density is given by  f =

∞

4π r 2 ρ(r)

0

=

4πρ0 (sR0 )3

sin sr dr = sr



R0

4π r 2 ρ0

0

sin sr dr sr

[sin (sR0 ) − sR0 cos (sR0 )]

(b) The intensity of scattering will be proportional to the square of the scattering factor I ∝ [sin (sR0 ) − sR0 cos (sR0 )]2 and will vary as a function of sR0 as shown in the following figure.

b3 0.569 0 0.2001 0.8694 0.2149 19.5424 20.2626 15.933 11.3966 23.4941 16.51

b4

c

51.651 0.216 0 1.594 14.039 0.404 85.341 1.2314 6.6537 0.7463 60.4486 16.378 98.74 1.1832 79.034 1.40818 64.8126 1.191 49.706 −2.567 76.9117 14.3992

OUP CORRECTED – FINAL

X-Ray Diffraction

I (s)

422

0

5

10 sR0

15

20

Further, the scattered intensity will be zero when sin (sR0 ) − sR0 cos (sR0 ) = 0 or tan (s0 R0 ) = s0 R0 . Thus, from successive minima in the scattering, which correspond to successive roots for s0 , we can determine R0 . Please note that often the symbol q is used in the scientific literature in place of s = sinλ θ .

7.4 Scattering by a Crystal: Structure Factor Now that we have described how the X-ray beam is scattered by an atom, our next step is to sum up the contributions from all the atoms in a crystal and specifically determine how they contribute to the diffracted intensity. Strictly speaking, as the X-ray beam propagates in a crystal it is attenuated due to successive scattering by the atoms. Moreover, beams “reflected” by atomic planes deep in the crystal can be re-reflected, and it is possible that these rereflected beams interfere destructively with the incident beam. As a result, one needs a “dynamical” theory to describe the true interaction of the X-ray beams in a crystal and predict the diffracted intensities. However, if the volume of the overall crystal is small, such dynamical effects can be ignored. Fortunately, this is the case for X-ray diffraction, and in practice a single-scattering model, also known as the kinematic theory, describes the intensities quite well. In contrast, for electron diffraction (§8), a dynamical theory is important. To calculate diffraction intensities, we will assume that Bragg’s law (§4.3.1) and the equivalent Laue condition (§4.3.2) are satisfied. Further, since the crystal can be considered to be a periodic repetition of the unit cell, we only need to determine the exact nature of the scattering from the atoms in the unit cell and their contributions to the diffracted intensity. We will also see that the intensities of the diffraction patterns depend on the nature—-centered or not—-of the unit cell, and the symmetry elements present in the crystal. In particular, symmetry elements, e.g. glide planes and screw axes, lead to systematic absences in intensities for diffraction from certain planes, even though they may satisfy the Bragg conditions. In such cases, the scattering from different atoms in the unit cell interfere destructively and result in zero observed intensities.

OUP CORRECTED – FINAL

Scattering by a Crystal: Structure Factor 423 We follow, qualitatively, the same approach as in the previous section for determining the scattering from a single atom. Instead of determining the phase difference for scattering from electrons, we calculate the phase differences and amplitudes for scattering from all the atoms in the unit cell. We will see that the problem of scattering from a unit cell resolves into the addition of waves with different amplitudes and phases, but of the same frequency, following either the vector method introduced in §6.2.3, or in the general case, using the complex notation presented in §6.2.4. The resultant wave, with scattering contributions from all the atoms in the unit cell, contributing to diffraction by a specific set of planes, (hkl), is then described by the structure factor, Fhkl , and just like the atomic scattering factor, fatom , (7.3.9), it can be defined as |Fhkl | =

amplitude of the wave scattered by all the atoms in a unit cell (7.4.1) amplitude of the wave scattered by a single electron

The corresponding diffraction intensity, Ihkl , is given by |Fhkl |2 , or strictly ∗ F , where F ∗ is the speaking, since it is a complex number, as Ihkl = Fhkl hkl hkl complex conjugate of Fhkl , with additional corrections for specific experimental factors to be described later in §7.10. To begin, consider diffraction from a primitive lattice, with only one atom, scattering factor, f 0 , at the origin (Fig. 7.4.1a). The incident and scattered X-ray beams satisfy the Bragg condition for the lattice planes (hkl) shown in the figure. Then, the beams scattered from all the atoms on the surface would be in phase, and by definition, the beams scattered from successive planes, dhkl , 2dhkl . . . ndhkl , would exhibit phase differences of 2π, 4π . . . 2n π. In other words, they would all be in phase, interfering constructively, to give a scattering amplitude that is a sum of the atomic scattering factor. Since we have only one atom in the unit cell, the structure factor for this trivial case is then Fhkl = f0 . Next, consider the same primitive lattice, but with two atoms, one at the origin with scattering factor, f 0 , and the other, with scattering factor, f 1 , at a position, r1 , with fractional coordinates (u1 , v1 , w1 ) in the unit cell (Fig. 7.4.1b). Thus r1 = u1 a + v1 b + w1 c

(7.4.2)

where, a, b, and c, are the unit cell vectors. Further, the incident beam, with unit vector, s0 , and diffracted beam, s, satisfy the diffraction condition for the lattice planes, (hkl), with inter-planar spacing, dhkl , as shown. The path difference (Fig. 7.4.1b) for the beams scattered from the two atoms is given by s − s0 ) P.D. = AC–BD = r1 · (

(7.4.3)

Since the diffraction satisfies the Laue condition k − k0 = ghkl =

2π s − s0 ) = ha∗ + kb∗ + lc∗ ( λ

(4.3.7)

OUP CORRECTED – FINAL

424

X-Ray Diffraction (a) θ

θ

θ f0

θ f0 dhhkl kl

θ

θ

θ

θ f0

f0 (b) f1 Figure 7.4.1 Bragg diffraction for lattice planes (hkl), from a primitive lattice with (a) one atom at the origin, and (b) two atoms with atomic scattering factors, f 0 and f 1 , located in the unit cell. Adapted from Hammond (2006).

θ

f0

θ

θ

θ f0

θ f1

θ

s D f1 B

f0

f1 θ

s0

f1

dhhkl kl θ

θ

A f0

f0

θ

C

Thus, we have s − s0 =

λ  ∗ ha + kb∗ + lc∗ 2π

(7.4.4)

and the path difference, (7.4.3), is now r1 · ( s − s0 ) =

λ  ∗ ha + kb∗ + lc∗ · (u1 a + v1 b + w1 c) = λ (u1 h + v1 k + w1 k) 2π (7.4.5)

because by definition a · a∗ = b · b∗ = c · c∗ = 2π , and a · b∗ = 0 etc.; see (4.2.4). Since a path difference of λ is equal to a phase difference of 2π, we can write the phase difference, φ 1 , between the waves scattered by the two atoms as φ1 = 2π (u1 h + v1 k + w1 k)

(7.4.6)

Following the graphical method for the superposition of waves, using the vectorphase diagram introduced in §6.2.3, where the vector lengths are proportional to the atomic scattering factors, f 0 and f 1 , with a phase angle, φ 1 , we obtain the resultant structure factor, Fhkl (Fig. 7.4.2a).

OUP CORRECTED – FINAL

Examples of Structure Factor Calculations 425

(a)

f3

(b)

φ3 f2 φ2

Fhkl Fhkl 

f1 φ1



f0

Figure 7.4.2 Graphical representation for the superposition of waves scattered by (a) two and (b) four atoms in the unit cell. The resultant wave has an amplitude given by the structure factor, Fhkl , and a phase difference, , with respect to the incident beam.

φ1

f1

f0

We can readily extend this analysis to all n atoms, ri , each with fractional coordinates (ui , vi , wi ), and scattering phase angles, φ i , occupying the unit cell as a summation. Since it becomes cumbersome to do this in vector form, as illustrated, even for four atoms (Fig. 7.4.2b), we employ the complex notation introduced in §6.2.4, and write the structure factor in general terms as Fhkl =

n  i=1

fi exp iφi =

n 

fi exp 2πi (ui h + vi k + wi k)

(7.4.7)

i=1

where the summation is over all atoms, n, in the unit cell. Since the structure factor is a complex number, the intensity of a particular diffraction peak is given by ∗ Fhkl Ihkl = Fhkl

(7.4.8)

We now demonstrate the calculation of the structure factor using various examples.

7.5 Examples of Structure Factor Calculations First it is important for the reader to review some important relations4 of complex exponential functions that we encounter in calculating structure factors.

7.5.1 Face-Centered Cubic (FCC) Structure The face-centered cubic unit cell (International Tables, Space group #225, Fm3m) shown in Figure 4.1.11f is characterized by a basis of four atoms, with fractional coordinates, uvw, given by 000, ½ ½ 0, ½ 0 ½, and 0 ½ ½. Assuming that each of these four positions are occupied by an atom of the same kind, with atomic scattering factor, f, we can write the structure factor as

4

eiφ = cos φ + i sin φ eiφ + e−iφ = 2 cos φ einπ = (−1)n = e−inπ

OUP CORRECTED – FINAL

426

X-Ray Diffraction Fhkl =

4  i=1

fi e2πi(ui h+vi k+wi l) 

= f e2πi.0 + e

2πi

h k 2+2





+e

2πi

k l 2+2





+e

2πi

l h 2+2



(7.5.1)

using the relationship einπ = (−1)n , we can rewrite (7.5.1) as   Fhkl = f 1 + (−1)h+k + (−1)k+l + (−1)l+h

(7.5.2)

It is now easy to see that when h, k, and l, are either all even or all odd, referred to from now on as unmixed indices, each of the four terms in Fhkl , (7.5.2), has a value equal to +1, and the structure factor becomes Fhkl = 4f

unmixed indices

(7.5.3a)

with Ihkl = |Fhkl |2 = 16 f 2 . On the other hand, if the indices, h, k, and l, are mixed, i.e. two odd and one even or two even and one odd, the sum of the three exponents equals –1 and the structure factor is Fhkl = 1 − 1 = 0

mixed indices

(7.5.3b)

with Ihkl = 0. ˚ calculate the structure factor Example 7.5.1: For FCC Cu, a0 = 3.61 A, ˚ for the first three allowed reflections, using Cu Kα radiation (λ = 1.54 A). Solution: FCC crystals show non-zero values of the structure factor for only unmixed indices. See Table 7.5.1. The first three allowed reflections are, (111), (200), and (220). From Table 4.1.2, for a cubic system dhkl = √ 2 a0 2 2 . h +k +l

The parameter s = sinλθhkl = 2d1hkl . We use the Cromer–Mann parameters for Cu, and calculate the scattering factors using (7.3.10). Finally, the structure factor for unmixed indices is given by (7.5.3a). We now solve in the form of a table. Reflection (111) (200) (220)

dhkl = √

a0 h2 +k2 +l 2

2.08 Å 1.805 Å 1.276 Å

shkl =

sin θhkl

λ

0.2404 0.277 0.3917

=

1 2dhkl

fhkl (s) =

4 

 ai exp −bi s2 + c

Fhkl = 4 f

i=1

22.04 20.71 16.77

88.19 82.84 67.07

OUP CORRECTED – FINAL

Examples of Structure Factor Calculations 427

7.5.2 Body-Centered Cubic (BCC) Structure The structure factor for a BCC structure (International Tables, Space group #229, I m3m) (Fig. 4.1.11i) can be calculated considering that the same kind of atom is located at the coordinates, 000, and ½ ½ ½, within the unit cell, as

Fhkl = f e

 2πi.0

+e

2πi

h k l 2+2+2



    = f 1 + eπi(h+k+l) = f 1 + (−1)h+k+l (7.5.4)

Thus Fhkl = 2f Ihkl = 4f 2

for h + k + l = even

(7.5.5a)

and Fhkl = 0 Ihkl = 0

for h + k + l = odd

(7.5.5b)

Note that, in addition, the structure factor is independent of the size and shape of the unit cell. Thus, any reflection with h + k + l = odd, will always be missing (= zero intensity) in a diffraction pattern for a body-centered cell, irrespective of whether it is cubic, orthorhombic, or tetragonal. The intensities based on the structure factors for diffraction of different planes (hkl) in FCC and BCC crystal structures are compared in Table 7.5.1. Note that many of them do not give rise to a diffracted beam with an observable intensity because of centering extinctions. When Bragg’s law is satisfied, we can use the interplanar spacing, dhkl (Table 4.1.2) for any crystal system, to determine the value of s, i.e. s = sinλθhkl = 2d1hkl , and using the appropriate atomic scattering factors from the Cromer–Mann parametrization, we can calculate the ideal values of the structure factors and the diffracted intensity. Table 7.5.1 Intensities for diffraction from different planes in BCC and FCC crystals. (hkl) BCC FCC

100

110

0 0

4f 2 0

111

200

0 16f 2

4f 2 16f 2

210

211

220

221

300

311

0 0

4f 2

4f 2

0 0

0 0

0 0

0

16f 2

Example 7.5.2: Calculate the structure factor for some select reflections ˚ for CrBCC using λ = 1.54 and lattice parameter a0 = 2.884 A. Solution: We solve this by applying (7.5.5), using values for the atomic scattering factor from Table 7.3.1, in the form of a table. Note that only h + k + l = even, have observable intensities.

222 4f 2 16f 2

OUP CORRECTED – FINAL

428

X-Ray Diffraction

sin θhkl

hkl

s=

000 100 011 111 200 210 121 120 310 311 320

0 0.173 0.245 0.3 0.347 0.388 0.425 0.388 0.5482 0.575 0.625

λ

=

1 2dhkl

=

√

h2 +k2 +l 2 ˚ −1 (A ) 2a0

f (Table 7.3.1)

FCr (7.5.5)

24 16.858 13.615 11.70 9.43 -

48 0 33.716 0 27.23 0 23.40 0 18.86 0 0

7.5.3 Hexagonal Close Packed (HCP) Structure We refer to the primitive hexagonal structure (Fig. 4.1.11e) with the unit cell containing two identical basis atoms, with atomic scattering factor, f, occupying positions with fractional coordinates, 000—the A layer, and 1/3 2/3 ½—the B layer. Substituting these two coordinates, we get the structure factor    

l l 2πi 3h + 2k 2πi 3h + 2k 3 +2 3 +2 =f 1+e Fhkl = f e2πi.0 + e

(7.5.6)

Let us consider two cases hkl = 0002 ∼ 00.2, and hkl = 1010 ∼ 10.0. We can see that F00.2 = f [1 + 1] = 2f

(7.5.7a)



  2πi 2π 2π F10.0 = f 1 + e 3 = f 1 + cos + i sin = f [0.5 + i0.866] (7.5.7b) 3 3 The structure factor for the latter is not real but complex, and to calculate the ∗ . We now rewrite intensity, we have to multiply Fhkl by its complex conjugate, Fhkl (7.5.6) as  

  l 2πi 3h + 2k 3 +2 = f 1 + e2πip Fhkl = f 1 + e

where p =

h 3

+

2k 3

+

l 2

(7.5.6)

to give, the general intensity

     Ihkl = f 2 1 + e2πip 1 + e−2πip = f 2 2 + e2πip + e−2πip = f 2 [2 + 2 cos 2πp]    = f 2 2 + 2 2cos2 π p − 1 = 4f 2 cos2 π p (7.5.7)

OUP CORRECTED – FINAL

Examples of Structure Factor Calculations 429 Expanding p, we get  Ihkl = 4f cos π 2

2

l h + 2k + 3 2

 (7.5.8)

l Now, we can readily see that Ihkl = 4f 2 , when h+2k 3 + 2 = n, where n is an integer. In addition, when h+ 2k = 3m, where m is an integer, and l is odd, we get Ihkl = 0. In practice, this leads to missing reflections, or absences, for various hkl such as 11.1, 11.3, 22.1, 22.3, etc.

7.5.4 Cesium Chloride (CsCl) Structure The structure is illustrated in Figure 4.1.13a, with Cl– located at 000 and Cs+ located at ½½½. Using atomic scattering factors,5 fCl and fCs , we get for the structure factor 

Fhkl = fCl e2πi.0 + fCs e

2πi

h k l 2+2+2



= fCl + fCs eπi(h+k+l) = fCl + fCs(−1) h+k+l (7.5.9)

Thus Fhkl = fCl + fCs for h + k + l = even Fhkl = fCl − fCs for h + k + l = odd

(7.5.10)

Note that in both cases in (7.5.10), the structure factor is a real number. We now have two intensities Ihkl = (fCl + fCs )2 for h + k + l = even Ihkl = (fCl − fCs )2 for h + k + l = odd

(7.5.11)

with the former intensities proportional to the square of the sum of the atomic scattering factors and called fundamental reflections, and the latter, substantially weaker reflections, with intensities proportional to the square of the difference of the atomic scattering factors, and called superlattice reflections. Further, if Cs+ and Cl– are randomly distributed in the same BCC structure, then they would randomly occupy the two positions 000 and ½½½, and the atomic Cs scattering factor for each site would be the average value, fCl +f 2 , and satisfying the symmetry requirements the reflections with h + k + l = odd would have zero intensity. In other words, the intensities would behave the same way as for a BCC crystal. However, if there were chemical order in the crystal, deviating from a random distribution of the two ions, then the superlattice reflection would show a finite intensity proportional to the degree of order, with maximum intensities observed for the ideal CsCl structure (see §7.11.4 for further details of chemical order–disorder transitions).

5 Strictly speaking, we should use the scattering factors for ions and not atoms, but here we neglect the difference.

OUP CORRECTED – FINAL

430

X-Ray Diffraction Example 7.5.3: The structure of sodium chloride, NaCl (Fig. 4.1.13b), can be interpreted as two interpenetrating FCC lattices, one occupied by Cl– , and the other by Na+ . The Na+ lattice is displaced with respect to the Cl– lattice by the vector, τ = a20 ( x + y + z), where a0 is the lattice parameter, giving atom locations Cl– Na+

000 ½½½

½½0 00 ½

½0½ 0½0

0½½ ½00

Calculate the structure factor, Fhkl , and intensity, Ihkl , for NaCl. Solution: Since both Cl– and Na+ occupy an FCC lattice, the structure factor for NaCl can be derived from that of the FCC structure, applying (7.5.2) to both the ions as   Fhkl = fCl e2πi.0 1 + (−1)h+k + (−1)k+l + (−1)l+h    2πi 2h + 2k + 2l + fNa e 1 + (−1)h+k + (−1)k+l + (−1)l+h    = fCl + fNa eπi(h+k+l) 1 + (−1)h+k + (−1)k+l + (−1)l+h    = fCl + fNa (−1)h+k+l 1 + (−1)h+k + (−1)k+l + (−1)l+k (7.5.12) The first term corresponds to the contribution of the basis atoms of Cl– at 000 and Na+ at ½½½, and the second term is the contribution from the face-centering translations already encountered in §7.5.1. Then, the intensities for the structure are given by  2  2 1 + (−1)h+k + (−1)k+l + (−1)l+h Ihkl = |Fhkl |2 = fCl + fNa (−1)h+k+l (7.5.13) Thus, similar to the FCC structure, Ihkl (NaCl) = 0, for mixed indices. However, for unmixed indices   Fhkl = 4 fCl + fNa (−1)h+k+l and

2  Ihkl = 16 fCl + fNa (−1)h+k+l = 16[fCl + fNa ]2 = 16[fCl − fNa ]2

for h + k + l = even for h + k + l = odd

Thus, we can clearly see that the addition of another atom to the basis has decreased the intensities of some of the reflections. This analysis can be readily extended to the diamond structure (see Problem 7.1).

OUP CORRECTED – FINAL

Symmetry and Structure Factor 431

7.6 Symmetry and Structure Factor 7.6.1 Crystals with Inversion Symmetry These crystals are also referred to as those with a center of symmetry at the origin. In these structures, for every atom at position, r, with fractional coordinates uvw, there exists another atom of the same kind at position, –r, with fractional coordinates uvw, in the unit cell. Then, the structure factor can be split into two terms Fhkl =

N 

fj e2πi (huj +kvj +lwj ) =

j=1

=2

N/2 

N/2 

fj e2πi (huvj +kvj +lwj ) +

j=1

N/2 

fj e−2πi (huj +kvj +lwj )

j=1

 fj cos 2π huj + kvj + lwj

(7.6.1)

j=1

Thus, we can conclude that the structure factor for a centrosymmetric crystal is always real for all reflections.

7.6.2 Friedel Law From the preceding, it follows that the diffraction pattern for centrosymmetric crystal6 is also centrosymmetric. Now consider the reflections Ihkl and Ihkl , i.e. two reflections that are on opposite sides of the direct beam, for a noncentrosymmetric crystal, in a primitive lattice with one atom at fractional coordinates uvw in the unit cell. Then, its structure factor, Fhkl = fe2πi(hu+kc+lw) , gives an intensity ∗ Ihkl = Fhkl Fhkl = f 2 e2πi(hu+kv+lw) e−2πi(hv+kv+lw)

= f 2 e2πi(hu+kv+lw) e2πi

 hu+kv+lw

= Ihkl

(7.6.2)

Thus, even if a crystal does NOT have a center of symmetry, the corresponding diffraction pattern will be centrosymmetric. This rule, known as Friedel law, has important consequences. Recall that there are only 32 point groups possible in 3D crystallography (§4.1.7). As a result of Friedel law, these 32 point groups are reduced to 11 centrosymmetric groups, called Laue groups, in their diffraction patterns. This is particularly relevant for electron diffraction, especially in a transmission electron microscope (see §8.5). Further details on Friedel law can be found in Hammond (2006) and Giacovazzo (2002).

7.6.3 Systematic Absences Certain symmetry elements, such as glide planes and screw axes, when present in a crystal structure give rise to systematic absences of certain specific reflections in diffraction. A glide plane is a symmetry element that simultaneously combines

6 Georges Friedel (1865–1933) was a French mineralogist and crystallographer.

OUP CORRECTED – FINAL

432

X-Ray Diffraction a reflection or mirror with a translation. Similarly, a screw axis is a rotation axis combined with a translation. We demonstrate the effect of a glide plane on the structure factor with the following example. Example 7.6.1: Consider an n-glide mirror, parallel to the (001) plane going through the origin with a translation component, τ = 12 ( x + y). Identify any systematic absences when this glide plane is present. Solution: When an n-glide is present, for each atom at position uvw, there is an atom of the same kind at u + 1/2, v + 1/2, w. The structure factor is then given by Fhkl = =

N/2  j=1 N/2 

e2πi (huj +kvj +lwj )

N/2 

      2πi h uj + 12 +k vj + 12 −lwj

+ fj e j=1   

  2πi h+k −wj l 2 fj e2πi (uj h+vj k) e2πiwj l + e fj

(7.6.3)

j=1

Then, for any reflection of the type, hk0, we get Fhk0 =

N/2 

   fj e2πi (uj h+vj k) 1 + eπi(h+k)

j=1

=

N/2  j=1

 fj e2πi (uj h+vj k) 2 cos π

   h + k πi h+k 2 e 2

(7.6.4)

It is easy to see that Fhk0 = 0, when h + k = 2m + 1, where m is an integer. To satisfy this criterion, if such an n-glide plane were to be present, there will be systematic absences of hk0 reflections, such as 210,120, 320, 430, . . . etc. Similar absences can be calculated for other types of glide planes or screw axes. More details can be found in de Graef and McHenry (2007), and in Problem 7.6.

7.7 The Inverse Problem of Determining Structure from Diffraction Intensities It is clear from §7.5 and §7.6 that, once the structure or the symmetry elements in the unit cell is known, in principle, we can calculate the diffraction intensities. In practice, we also need to correct them for the experimental factors, discussed in §7.10, that influence the diffraction intensities. However, because we are only measuring the intensities, we are unable to say anything meaningful about the phase, , of the diffracted wave. Unfortunately, this phase information

OUP CORRECTED – FINAL

Broadening of Diffracted Beams and Reciprocal Lattice Points

433

is critical to solve the inverse problem that is often of interest. Note that all vectors, Fhkl , such as in Figure 7.4.2b, as long as they have the same modulus will give the ∗ F , irrespective of their phase,  , or the direction in same intensity, Ihkl = Fhkl hkl hkl which Fhkl points. In other words, if we are given an unknown crystal structure, we need both the structure factor, Fhkl , and the phase, hkl , of the diffracted wave for several sets of planes, (hkl), to unequivocally identify the positions of the atoms. In special cases, such as the centrosymmetric crystal with the origin coinciding with the center of symmetry (§7.6.1), where the structure factor, Fhkl , is a real number with no imaginary component, the inverse problem can be solved. However, in general for non-centrosymmetric crystals, the inverse problem cannot be solved but close approximation can be made using advance techniques that are beyond the scope of this book, but can be found in specialized texts, e.g. Giacovazzo et al. (2002).

7.8 Broadening of Diffracted Beams and Reciprocal Lattice Points

Intensity (a.u.)

Ideally, in a diffraction experiment satisfying Bragg’s law, λ = 2dhkl sin θhkl , the diffracted beam should appear as a sharp line with intensity, Ihkl = |Fhkl |2 , at a position, 2θhkl , defined by the relation between the diffracted and the incident or primary beam (Fig. 7.8.1). In practice, for crystals of finite thickness, the diffracted beam is broadened (Fig. 7.8.1) and its intensity is modified by experimental factors as discussed in §7.10. The broadening is conceptually similar to that discussed for Fraunhofer diffraction (§6.6.2) from a grating with a finite number of slits. To derive an expression for the broadening of the diffracted beam, we consider a finite crystal with thickness, t = mdhkl , perpendicular to the set of diffracting

2θ = β

Ideal 2(θhkl – θ)

2θhkl

Experimental 2(θhkl + θ)

2θ

Figure 7.8.1 Broadening, β, of a diffraction peak can be due to a variety of factors including the crystallite or grain size, microstrain, and instrument factors (see §7.11.2).

OUP CORRECTED – FINAL

434

X-Ray Diffraction A0

b0

b0

A1

b1

b1

A1

A2

b2

b2

A2

0

A0

dhkl

1 Figure 7.8.2 Bragg reflection from a crystal of thickness, t, measured perpendicular to the set of planes with spacing, dhkl . Arrows represent  the incident (An ), and reflected An beams at the exact Bragg orientation, θhkl , for the set of successive planes  n = 0, . . . , m; similarly, bn , bn represent incident and reflected beams at a small deviation, θ , from the Bragg angle. Adapted from Hammond (2006).

2

bm/2 A m/2

Am/2 bm/2

m/2

t = mdhkl

Am

bm

θ θhkl

bm

θ

Am

θhkl

m

planes, dhkl (Fig. 7.8.2). The planes satisfy the exact diffraction condition at the Bragg angle, θhkl , for the incident beams, A0 , A1 , . . . . Am/2 . . . . Am , with diffracted beams, A0 , A1 , . . . . Am/2 . . . . Am , reflected from successive planes, 0 . . . m/2 . . . m (Fig. 7.8.2). Since Bragg’s law is satisfied, the path difference between rays reflected from plane, 0, and successive planes in the crystal is nλ, where n = 1, . . . , m/2. . . . ..m; this leads to constructive interference for rays reflected from all planes, as expected for Bragg diffraction. It is impractical to obtain a truly parallel primary beam, in spite of our best efforts at collimation (see §7.9.5.1), and a small angular deviation, θ , of the incident beam is always to be expected. Now, consider the set, b0 , b1 , . . . . bm/2 . . . . bm , of such primary beams, all with angular deviation, θ , incident on successive planes with reflected beams, b0 , b1 , . . . . bm/2 . . . . bm , also shown in the figure. Since θ θhkl , the path difference between b0 b0 and b1 b1 is insignificantly different from λ, and a constructive interference between the two will persist. However, for consecutive planes, n, the path difference (P.D.) between b0 b0 and bn bn , will gradually increase with n, and at some critical depth, the path difference will be such that it will give rise to a destructive interference, i.e. P.D. = λ/2. Let the first two such planes that show constructive interference be n = 0 and n = m/2; they then satisfy Bragg’s law m m λ = 2 dhkl sin θhkl 2 2

(7.8.1)

OUP CORRECTED – FINAL

Broadening of Diffracted Beams and Reciprocal Lattice Points

435

For destructive interference from the same set of planes, at small angular deviation, θhkl + θ , we get from Bragg’s law m λ m λ + = dhkl sin (θhkl + θ) 2 2 2

(7.8.2)

When this condition, (7.8.2), is satisfied all subsequent pairs of lattice planes, 1 and m/2 + 1, 2 and m/2+2, . . . , also satisfy the destructive interference criterion, such that the entire crystal as a whole will result in a destructive interference. When this happens, we expect to see zero intensity in the diffracted beam, thus defining the width of the diffracted beam on either side of the Bragg angle, θhkl . We can expand (7.8.2) to get m λ m λ + = 2 dhkl [sin θhkl cosθ + cos θhkl sin θ ] 2 2 2

(7.8.3)

For small angles, sinθ = θ , cosθ = 1, t = mλ, and using (7.8.1), we can simplify (7.8.3) as λ = t cos θhkl θ 2

(7.8.4)

or 2θ =

λ =β t cos θhkl

(7.8.5)

The relationship (7.8.5) that relates the broadening, β, of the diffracted X-ray beam to the thickness, t, of the crystal in a direction normal to the lattice planes, dhkl , is known as the Scherrer 7 equation. Here, β is the angular width of the beam at half the maximum peak height, as defined in Figure 7.8.1. The Scherrer equation (7.8.5) is of great practical utility as it relates the peak broadening to the average crystallite or grain size in a specimen (see §7.11.2). However, in practice, to accommodate the overall shape of the crystal (7.8.5) is modified as β=K

λ t cos θhkl

(7.8.6)

where K (∼1) is an empirical constant that accounts for the crystallite shape. It is also possible to include this broadening in the Ewald sphere construction (Fig. 4.3.2) by extending the infinitesimal reciprocal lattice point to form a “node” of finite size (Fig. 7.8.3). Now it can account for the fact that the diffracted beam satisfies the Laue condition (4.3.7), even when it is broadened by the angular range of 2β = 4θ . This is possible if and only if the reciprocal lattice point is broadened to a “node” of finite dimensions/length allowing it to intersect the reflecting sphere as the crystal rotates. Let ghkl represent the extension of the reciprocal lattice point about its mean position.

7 Paul Hermann Scherrer (1890–1969) was a Swiss physicist and the co-inventor of the Debye–Scherrer method (§7.9.3).

OUP CORRECTED – FINAL

436

X-Ray Diffraction Reciprocal lattice “node” 2β k

2θhkl

Incident X-ray beam

Figure 7.8.3 The Ewald sphere construction for a reflected beam broadened by 2β corresponding to the extension of the reciprocal lattice node by 2π/t, where t is the thickness of the crystal in a direction normal to the diffracting planes.

ghkl

k0

Ewald sphere

Since | ghkl |= ghkl =

2π 4π sin θhkl = dhkl λ

(7.8.7)

we get  ghkl = 

4π sin θ λ

 =

4π cos θ θ λ

(7.8.8)

and substituting for θ from (7.8.5), we get ghkl =

4π cos θ λ 2π = λ 2t cos θ t

(7.8.9)

Thus, the extension of the reciprocal lattice point is proportional (by a factor of 2π) to the inverse of the crystal dimension (thickness, t) in a direction normal to the diffracting planes. We can now generalize this argument to all other directions of the crystal and say that the reciprocal lattice nodes have a finite size that is inversely related to the shape and size of the crystal. For thin films, surfaces, or plate-like crystallites, the nodes become rods or streaks in a direction normal to the thickness. This is particularly important in electron diffraction in a transmission electron microscope (see Fig. 8.5.4) where electron transparent thin foils of specimens are used, and where the corresponding reciprocal lattice nodes are referred to as rel-rod (see §8 and §9). Further, as we have seen in Fraunhofer diffraction for a finite number of gratings (§6.6.2), the nodes in X-ray diffraction are also surrounded by subsidiary minima and maxima. If the crystal thickness

OUP CORRECTED – FINAL

Methods of X-Ray Diffraction 437 is large, these maxima/minima can be ignored in X-ray diffraction. However, in thin film multilayers, where there are a finite number of layers, these secondary maxima and minima are also observed and can provide additional information about the specimen; this is discussed further in §7.9.4, and specifically in Figure 7.9.12. Example 7.8.1: Consider a powder diffraction experiment of polycrystalline iron using Cr Kα X-ray radiation. (a) What would be the first three most intense reflections? (b) What would be the peak broadening in degrees for these three reflections if the crystallites were (i) 1 mm and (ii) 50 nm, in diameter. Comment on the results. Solution: (a) We assume BCC iron with lattice parameter, a = 0.2866 nm. From Table 7.5.2, we determine that the three most intense reflections are (011), (002), and (211). (b) The wavelength of Cr Kα is λ = 0.2291 nm. We apply Bragg’s law to determine the angle of diffraction, and (7.8.6) to determine peak broadening (assuming K ∼1), and solve in the form of a table. hkl

d hkl (Å)

θ hkl

β(rad) d = 1mm

β(deg) d = 1mm

β(rad) β(deg) d = 50 nm d = 50 nm

011 2.027 34.41◦ 2.78 × 10–7 1.6 × 10–5 0.0056 200 1.433 53.07◦ 3.81 × 10–7 2.2 × 10–5 0.0076 211 1.17 78.25◦ 1.125 × 10–6 1.0 × 10–4 0.0225

0.32◦ 0.44◦ 1.29◦

Clearly, for the 1 mm crystallites, all reflections have negligible broadening. However, for the 50 nm crystallites, we see that the diffraction peaks would be rather broad with a measurable broadening of 0.32◦ –1.29◦ . Further, the broadening increases with the Bragg angle or smaller interplanar distances.

7.9 Methods of X-Ray Diffraction Bragg’s law (4.3.3) and its equivalent formulation in reciprocal space, the Ewald sphere construction (Fig. 4.3.2), and the Laue criterion (4.3.7) are the basis for the applications of X-ray diffraction in materials characterization. We have already seen the use of Bragg’s law in X-ray analysis, especially in the form of wavelength

OUP CORRECTED – FINAL

438

X-Ray Diffraction dispersion spectroscopy (§2.5.1). In that case, using crystals of known inter-planar spacing, dhkl , the intensities as a function of scattering angle, θ, were measured, and when Bragg’s law is satisfied, the unknown wavelength, λ, is determined. Here, our interest is to use X-ray diffraction to determine the lattice parameter and structure of an unknown material, be it in crystalline, polycrystalline or amorphous form. In general, there are two variables, θ and λ, that can satisfy Bragg’s law for a crystal with a series of values of interplanar spacing, dhkl . In a scattering experiment, for a given dhkl , we can satisfy Bragg’s law by either keeping θ fixed and varying λ, or keeping λ fixed and varying θ. The former, the Laue method, is traditionally applied to the study of single crystal specimens. The latter method, has a number of variations; two of them–the Debye–Scherrer method (fix λ and vary θ ), with particular applicability to the study of powder specimens, and diffractometry (fix λ and rotate the crystal/specimen), are discussed here. Note that we use the Ewald sphere construction and the Laue criterion to discuss these methods; before we proceed, it would be helpful to review §4.3.

7.9.1 The Laue Method for Single Crystals The geometry of the Laue method is straight forward and illustrated in Figure 7.9.1. A narrow beam of white radiation, such as the continuous spectrum from an X-ray tube (§2.4), is allowed to be incident on a fixed single crystal specimen. In the Laue method, either the forward diffracted beam is recorded in transmission (Fig. 7.9.1a), provided the specimen is sufficiently thin, or in the case of a bulk crystal, the backward diffracted beams are recorded by back reflection (Fig. 7.9.1b). In both cases an array of spots, or a diffraction pattern, is recorded (Fig. 7.9.1c,d). The common practice is to record the back-reflected pattern, as this allows the use of thicker, opaque specimens without the need for any specimen thinning procedures. Details of Laue X-ray diffraction can be understood in terms of the reflecting or Ewald sphere construction (Fig. 7.9.2), introduced in §4.3.2. The incident, Xray white radiation used in the experiment, can be considered to vary continuously from wavelengths λmin to λmax. In the figure, only two intermediate wavelengths, λ1 and λ2 , such that λmin < λ1 < λ2 < λmax , are explicitly shown. Thus, only the reciprocal lattice points in the shaded area (that would be intersected by the spheres of one of the many Ewald spheres at intermediate wavelengths) contribute to the observable diffraction pattern. Hence, for the case shown in Figure 7.9.2, 44 nodes contribute to the Laue pattern. In general, the Laue diffraction pattern produced in the laboratory is not easy to interpret for anything other than single crystal specimens. However, the advent of synchrotron radiation (§2.4.3) and powerful computers for interpretation have provided new avenues of application for the technique. In particular, the Laue method has been applied to resolve the structure of small molecules [4] and macromolecular crystallography [5], including dynamic, time-resolved measurements [6]. Finally, the simplicity of the experimental set-up makes for

OUP CORRECTED – FINAL

Methods of X-Ray Diffraction 439 (a)

(b)

(c)

Figure 7.9.1 The Laue method for single crystals in (a) the transmission, and (b) back-reflection, geometries. Examples of the back-reflection Laue pattern of an aluminum crystal with incident beam parallel to (c) [011] (note: shadow is from the goniometer holding the specimen), and (d) [001] clearly showing the fourfold symmetry of the intersecting zones.

(d)

Adapted from (c) Cullity (1978), and (d) Hammond (2006).

000

2π λmin

2π 2π λ2 λ1

2π λmax

a very portable Laue diffractometer, including remotely carrying out diffraction analysis of geological specimens on Mars (see §7.11.5).

7.9.2 Diffractometry of Powders and Single Crystals Consider a powder specimen with millions of grains aligned in random orientations, each of which contains various lattice planes, (hkl), that under appropriate conditions of diffraction angle, 2θhkl , satisfy Bragg’s law. In practice, as the direction of the incident beam is varied, there are two ways that the angle between the incident beam and the scattered beam can always be maintained at 2θ in a

Figure 7.9.2 The Ewald sphere construction using “white” radiation over a range of X-ray wavelengths, λmin < λ < λmax , for diffraction from a single crystal specimen. All the nodes (total = 44) in the shaded region are observed in the Laue pattern, but only a few (six) are indicated by arrows.

OUP CORRECTED – FINAL

2θ θ

r cto

cto Specimen stage

X-ray tube (stationary)

ω

(b) Stationary specimen stage

ω X-ray tube

ω Equitorial plane

De te

2ω

(a)

r

X-Ray Diffraction

De te

440

θ

θ Equitorial plane

Figure 7.9.3 (a) The Bragg–Bretano geometry, also known as the θ –2θ scan in X-ray diffraction. Here the specimen is rotated at half the angular velocity of the detector, and the source is stationary. (b) A θ–θ scan; here the specimen is stationary and both the source and detector are rotated at the same angular velocity, ω, but in opposite directions. diffractometer: (i) the X-ray tube or source is stationary and the detector and specimen rotate with angular velocities of 2ω and ω, respectively, in the same manner (Fig. 7.9.3a); this is known as the Bragg–Bretano geometry or commonly as the θ − 2θ scan, and (ii) the specimen is stationary and both the source and detector are rotated towards each other at the same angular velocity (Fig. 7.9.3b); this is known as the θ − θ scan. Both these arrangements, where the orientation of the X-ray source, the specimen, and the detector are well defined, ensure that Bragg’s law is satisfied at all times. Now, consider the θ − 2θ diffractometry from the reciprocal space point of view (Fig. 7.9.4). From the Laue criterion (4.3.7), we expect to observe a diffracted beam with sufficient intensity when any reciprocal lattice point intersects the surface of the Ewald sphere. In a powder specimen, the grains can be considered to be randomly oriented. Hence, its reciprocal lattice can be derived from that of the equivalent single crystal by the free rotation about the center of the limiting sphere. Thus, each reciprocal lattice point or node will generate a spherical surface, which will then interact with the Ewald surface in the form of a cone of allowed reflections, for all individual grains that subtend an angle, 2θ, with the incident beam. A diffractometer uses the equatorial geometry (Fig. 7.9.3), whereby the diffracted beam angles are measured in a plane defined by the incident X-ray beam and the rotation of the detector about an axis passing through the specimen as shown. As the angular relations between the detector, the polycrystalline powder specimen and the incident beam are varied systematically in the equatorial plane, either in the form of θ − 2θ or θ – θ scans, peaks corresponding to the various planes satisfying Bragg’s law are observed. Figure 7.9.5 shows a typical example of a powder specimen of NaCl. Note that, in practice, a diffractometer detector will not detect X-rays from planes where normal does not lie in the equatorial plane,

OUP CORRECTED – FINAL

Methods of X-Ray Diffraction 441 Reciprocal Lattice Points or “Nodes”

Incident X-Ray Beam Limiting Sphere

Ewald Sphere

Intensity (counts)

6000

200

4000 220

2000 222 311

111 0

Figure 7.9.4 Diffraction from a powder or polycrystalline specimen is equivalent to rotating the reciprocal lattice about the center of the limiting sphere. Each reciprocal lattice vector now forms a sphere, which then intersects the Ewald sphere in a cone of allowed reflections, representing all the individual crystallites/grains that subtend an angle, 2θ hkl , with the incident beam.

30

40

50 60 2θ (degrees)

400

420 331 70

422 80

even though they may satisfy the Bragg condition. In other words, the intersection of the Ewald sphere and the reciprocal lattice (sphere) is in the form of a cone (as shown for one reflection in Fig. 7.9.4), and we only measure one section in the equatorial plane. Further, only a small fraction of grains in the powder specimen will contribute to the observed pattern, as only planes with normal in the equatorial plane contribute to the diffracted intensities. As a result, it is necessary that the average powder size in the specimen be sufficiently small to give a statistically good sampling of all the possible lattice parameters in the specimen. Large databases, called powder diffraction files (PDFs), for data sets with extensive entries numbering well over 200,000 different materials, are published by the International Center for Diffraction Data.8 To identify any material, the 2θhkl

Figure 7.9.5 Powder diffraction data for NaCl. Notice that the h + k + l = even reflection have higher intensity than those with h+k+l = odd, consistent with the scattering factors for Na and Cl being in phase for the former and out of phase for the latter (see Example 7.11.1). Notice that the most intense reflections are 200, 220, and 222, in good agreement with powder data file shown in Figure 7.11.1.

8

http://www.icdd.com

OUP CORRECTED – FINAL

442

X-Ray Diffraction values for the first three most intense X-ray diffraction peaks are compared with these tables, and generally, this is the first step in solving the crystal structure and phase of a new or unknown material (see §7.11.3). Even though we have discussed powder diffractometers, single-crystal specimens may also be examined in a diffractometer by using a Eulerian cradle, also known as a four-circle goniometer, shown schematically in Figure 7.9.6, to mount and orient the specimen. As the name implies, there are four rotational motions available, three of which are associated with the crystalline specimen, and one with the detector/counter. The instrument has a main axis, normal to the equatorial plane containing the incident and diffracted beams passing through the crystal. There are two possible rotations about this main axis: the rotation of the detector, 2θ, to satisfy the Bragg diffraction condition for a particular reflection, and the rotation of the cradle, ω, as shown. Further, once the detector has been set at the Bragg angle, the scattering vector, g, can be oriented to make the correct angle with the incident and diffracted beams by rotating the crystal about two additional axes,  and χ, also shown in the figure. Note that, if the cradle is fixed with its χ -plane perpendicular to the incident X-ray beam, the ω degree of freedom is redundant; such a set-up is also common and is known as a three-circle goniometer. In practice, only two rotations are required to bring a reciprocal lattice node to the intersection of the Ewald sphere and the equatorial plane. Starting with χ = 0, a rotation about  (or ω) can bring the required reciprocal lattice node on the Ewald sphere as shown in Figure 7.9.7a. A subsequent rotation, χ , about the χ-axis can bring the node to the equatorial plane (Fig. 7.9.7b). Then, the diffracted beam can be measured by placing the detector at the required 2θ angle. Note that, in such arrangements, it is possible that the physical χ -circle may obstruct the scattered beam from reaching the detector; thus, the extra degree of  Rotation χ Rotation Beam trap

Figure 7.9.6 A four-circle diffractometer, also known as a Eulerian cradle, used for single crystal analysis. The detector rotates about the 2θ axis in the equatorial plane, and the specimen can be oriented in any way by a combination of three different rotation axes, ω, χ, and . Adapted from Woolfson (1997).

Incident X-ray beam

Specimen

2θ Detector Diffracted beam

ω Rotation

2θ Rotation

OUP CORRECTED – FINAL

Methods of X-Ray Diffraction 443

(a)

P

P ghkl

Incident X-ray beam

ghkl O

A

P

(b)

χP ghkl O 2θ

A Incident X-ray beam

freedom afforded by the four-circle goniometer can prove to be of valuable utility. Most modern single-crystal diffractometers with four-circle Eulerian cradles are computer controlled with sophisticated software to carry out the tasks of crystal orientation, data collection, and solving the crystal structure. Finally, both powder and single-crystal diffractometers can be modified with heating or cooling units to carry out measurements at both high and low temperatures (see §7.11.4).

7.9.3 Debye–Scherrer Method for Powders An earlier alternative to the powder diffractometer is the Debye–Scherrer camera that belongs to a family of transmission “cameras”, where the specimen is placed in between the X-ray source and a stationary recording film. In this family, the Debye–Scherrer camera (Fig. 7.9.8a) is unique, because even for standard specimen to film distances of ∼8.0 cm, it records a wide range of 2θ angles as it registers both the forward- and back-scattered X-ray beams. It also has the distinct advantage of working with very small quantities of powder specimens, that consists of a large number of tiny crystallites—-typical volume of crystallite is 5 μm3 and thus in a ∼1 mm3 specimen, we expect ∼107 crystallites—-arranged in a random orientation. The powder specimen is usually formed in the form of a needle-shaped cylinder, and housed in a thin-walled glass tube at the center of the camera. When a monochromatic, well-collimated X-ray beam is incident on the specimen in a Debye–Scherrer camera, the diffraction geometry is similar to that for the

Figure 7.9.7 Typical method to bring a reciprocal lattice node to the intersection of the Ewald sphere and the equatorial plane. (a) First, the crystal is rotated about  until the node lies on the sphere, as shown in this projection looking down the instrument main axis. (b) Second, it is rotated about χ to bring the node to the equatorial plane, where it can be measured by diffraction as shown. Adapted from Giagovazzo et al. (1992).

OUP CORRECTED – FINAL

444

X-Ray Diffraction (a)

(b)

Film

c Spe

Beam stop

Incident X-ray beam

Incident X-ray beam

Specimen Film

Direct beam

n ime

Diffracted beams from different crystallites

(c) A B

xhkl

2θ = 180°

L

2θ = 0°

Figure 7.9.8 (a) The Debye–Scherrer camera, showing the incident beam, the beam stop, the specimen, the film, and the beams diffracted in all directions. (b) The locus of the diffracted beams from all the crystallites in a powder specimen for a given reflection, hkl. (c) A straightened photographic film, showing the rings that originate from intersection of the cone in (b); the position of the ring, xhkl , normalized by the separation, L, of the incident, A, and beam-stop, B, positions give the angle θ hkl . (a) Adapted from Giagovazzo et al. (1992). (b) Adapted from Woolfson (1997). (c) Adapted from de Graef and McHenry (2007).

diffractometer (Fig. 7.9.4) with the difference being the specimen is cylindrical instead of a flat thin film. For the particular crystal structure in such a powder specimen, each possible crystallographic plane, (hkl), that satisfies the Bragg condition, 2θhkl , is now present in large numbers at random orientations because of the small crystallite size. As a result, all the diffracted beams for the specific reflection, hkl, make an angle, 2θhkl , with the incident beam and collectively form a cone of semi-angle, 2θhkl , originating at the specimen as shown in Figure 7.9.8b. When each cone of diffracted beams for a specific reflection intersects the circular film mounted on the inside of the camera, a diffraction pattern of ring-segments is recorded; when the film is straightened out, it appears as in Figure 7.9.8c. The two holes on the film correspond to the position (A) of the incident beam, or 2θ = 180◦ , and position (B) of the transmitted beam stop, or 2θ = 0◦ ; the

OUP CORRECTED – FINAL

Methods of X-Ray Diffraction 445 distance, L, between points A and B on the film is recorded. The value of 2θhkl for any particular diffraction ring can be determined by measuring its distance, xhkl , from position, B, and taking the ratio to give θhkl =

180◦ xhkl 2 L

(7.9.1)

Using Bragg’s law, each observed ring can be identified with a specific interplanar spacing, dhkl . Comparing the list of observed spacing with the PDF (§7.11.3), the structure may be identified. However, a large number of diffracted beams will have similar 2θ values causing them to overlap making it difficult to resolve their intensities separately (see Exercise 7.4). As a result, a diffractometer is preferred for the analysis of powder specimens; moreover, for new and unknown materials, generally X-ray diffraction of single crystal specimens gives the best structural analysis (see §7.11.5). If this is not possible, sophisticated refinement (§7.11.3) of powder diffraction data gives good results.

7.9.4 Thin Films and Multilayers: Diffractometry, Reflectivity, and Pole Figures As we have seen in §7.9.2 and in Figure 4.3.5), the standard or symmetrical Bragg– Bretano geometry (Fig. 7.9.3) provides the interplanar spacing of only those lattices planes parallel to the surface of a thin film specimen. This is achieved by orienting the incident and reflected beams, including all the required collimation, such that they make equal angles with the specimen surface (Fig. 7.9.9a). Figure 7.9.3 shows how, in practice, this geometry can be implemented in two ways by keeping the X-ray tube/source stationary and rotating both the specimen and detector, or alternatively, keeping the specimen/film stationary and rotating the source and detector in opposite direction but at the same angular rate. In this case, if the specimen is a single crystal film there will be only one peak; however, if the film is multilayer, such as the tri-layer film shown in Figure 7.9.10, there will be one (family of) peak(s) for each of the layers, and the substrate. Figure 7.9.9b shows an alternative asymmetric arrangement, common in X-ray diffractometry of thin films. Here, the specimen is tilted at an angle, α, which can be varied, but the angle between the incident and reflected beam is always maintained at 2θ. As a result, the incident, (θ + α), and reflected, (θ – α), beams make different angles with the specimen surface, thereby making it possible to record diffraction from planes that are NOT parallel to the specimen surface. Note that the maximum angle of specimen tilt, α = θ , and using the Ewald sphere construction (Fig. 7.9.9c) we can now see the regions of reciprocal space, or the portion of the limiting sphere that can be sampled by the symmetric and asymmetric scattering geometries.

OUP CORRECTED – FINAL

446

X-Ray Diffraction (a)

(b) ghkl nˆ

ghkl

Figure 7.9.9 Diffractometer settings for single crystals. (a) A symmetric setting where all the reciprocal lattice vectors involved in diffraction are from planes with normal to the specimen surface. (b) An asymmetric setting where the reciprocal lattice vectors are inclined by an angle, α, with respect to the surface normal, n. (c) All the possible reflecting planes lie within the shaded region of the limiting sphere.

θ+α

nˆ

θ

θ α dhkl

dhkl 4π λ

Maximum specimen tilt for reflection hkl

(c) θ

4π λ

θ

Specimen surface plane

4π λ MnPd(002)

MgO(002)

Adapted from [7].

Aucap(20Å) MnPd(330Å) Fe(80Å) MgO

Fe(002)

MnPd(001)

Intensity log (arb. units)

Figure 7.9.10 High angle, θ − 2θ X-ray scan, using Cu Kα radiation, from the multilayer structure (shown on the left), grown at different temperatures: (a) 100 ◦ C, (b) 200 ◦ C, (c) 350 ◦ C, and (d) 450 ◦ C. The orientation of the tetragonal MnPd film changes from an a-axis normal to a c-axis normal when the growth temperature is increased from 100 ◦ C to greater than 450 ◦ C.

θ–α

(d) MnPd(200)

(c) MnPd(002)

(b) (a) 20

30

MnPd(200)

40 50 2θ (deg)

60

70

We now present three distinct examples of diffraction from (1) a polycrystalline film, (2) a single crystal film subject to a -scan, and (3) a multilayer film at both small and large angles. (1) A polycrystalline film with many grains oriented in random directions, is mounted on a four-circle goniometer (Fig. 7.9.6) and aligned such that a specific reflection, hkl, is in the Bragg orientation, with a peak at 2θhkl observed in X-ray diffraction. This implies that some of the polycrystalline grains are oriented in such a way that they satisfy the Bragg condition. Now slowly rotate the crystal

OUP CORRECTED – FINAL

Methods of X-Ray Diffraction 447 around the -axis while maintaining the θ − 2θ scattering geometry in the equatorial plane. Since the polycrystalline film is composed of small crystalline grains oriented at random, we expect to continue seeing a diffracted intensity from some of the grains, independent of the value of . In other words, if we plot the intensity of the specific reflection, Ihkl , it should have the appearance of a circle, i.e. is the same for all values of . (2) Now consider a single crystal film, grown on another single crystal substrate. Again, set the scattering condition to satisfy Bragg’s law for a specific reflection, hkl. The multiplicity and the angles between the families of planes {hkl} is well defined for a crystal with a given symmetry; thus, rotating the crystal about the -axis will result in measurable intensity, Ihkl , only at those values of  consistent with the symmetry of the crystal. Such a plot of intensities for a given reflection, hkl, as both  and χ are varied is commonly referred to as a pole figure. We now illustrate this with a specific example. Figure 7.9.11 shows pole figures for MgO films grown epitaxially on single crystal substrates of (a) SrTiO3 (STO), a perovskite structure with a ∼3.90 Å, and (b) LaAlO3 (LAO), a perovskite structure with a ∼3.8 Å. Both pole figures were measured for the MgO222 reflection set at 2θ222 = 78.6◦ . Note that in these plots,  is the angle by which the specimen is rotated about the surface normal, and χ is the angle by which the surface is tilted out of the equatorial plane. Thus, the three concentric circles represent χ values of 30◦ , 60◦ , and 90◦ . In both figures, notice a ring of four MgO222 spots at χ ∼55◦ ; further, there is an

(a) MgO/SrTiO3

(b) MgO/LaAlO3



MgO

STO



MgO

LAO



Figure 7.9.11 Pole figure measurements showing the variation of  around the circumference of the circle, and χ increasing radially, with three concentric circles representing χ = 30◦ , 60◦ , and 90◦ , respectively. (a) For MgO/SrTiO3 , the MgO222 reflection at 2θ222 = 78.6◦ (♦), shows four peaks at χ = 55◦ , consistent with the epitaxial relationship, MgO 100 SrTiO3 100, as illustrated here. The STO(013) reflections at χ = 18◦ and 72◦ are circled. (b) For MgO/ LaAlO3 , the MgO222 reflections occur at the same  angles, illustrating a different epitaxial relationship rotated by 45◦ , i.e. MgO

100 LAO 110. Adapted from [7].

OUP CORRECTED – FINAL

448

X-Ray Diffraction inner (χ ∼ 18◦ ) and outer ring (χ ∼ 72◦ ) of spots that correspond to the substrate (013) reflections. However, there is one significant difference between the two figures. For the SrTiO3 substrate (Fig. 7.9.11a), the MgO222 peaks occur midway in  angles between the STO013 peaks, whereas for the LaAlO3 substrates, the MgO222 peaks occur at the same  angles as the LAO013 peaks. From this we can conclude [8] that the orientation relationship (Fig. 7.9.11) between the MgO films isMgO 100 SrTiO3 100, and MgO 100 LAO 110, and the relative lattice orientations for each case is as shown. (3) Multilayers are sequences of very thin crystalline films of two or more materials (such as Co and Pt), grown (often) on single crystal substrates. Each of the distinct layers displays their own characteristic high-angle Bragg peak. Typically, as the layers are very thin, these Bragg peaks are broadened, and in principle, it is possible to use their broadening to estimate the thickness of the individual layers following the Scherrer equation (7.8.5). However, the repeat distance, , of the multilayer or superlattice can also be determined from the location of the satellite reflections which occur on either side of the Bragg peaks. For a multilayer with long-range structural coherence, the repeat distance is given by =

λ 2 (sin θ2 − sin θ1 )

(7.9.2)

where λ is the wavelength of the X-rays, and 2θ 1 , 2θ 2 are the Bragg angles of adjacent satellite peaks. Converting the Bragg angles to nominal distances, we can rewrite (7.9.2) as 1 1 1 = −  d2 d1

(7.9.3)

grown Figure 7.9.12a shows an example [9] of a (Co3Å Pt15Å  )30 multilayer  ˚ radiation. on GaAs(111) with a Ag200Å buffer layer, using Cu Kα λ = 1.54 A Peaks are observed at 2θ values of 30.5 (Pt, n = −2), 35.5 (Pt, n = −1), 38.2 (Ag), 40.6 (Pt, n = 0), and 45.6 (Pt, n = 1). Using (7.9.2), it is easy to see that ˚ (see Problem 7). the multilayer period,  ∼ 18A At low angles (2θ = 1 − 5◦ ), we see additional Bragg reflections from the repeat period, , of the multilayers, i.e. reflections that satisfy λ = 2sinθ. These measurements at low angles, generally termed as reflectivity measurements, show very fine oscillations that depend on the total thickness, T, of the multilayer film, where T = n and n is the number of repeating units. The analysis of these fine oscillations is beyond the scope of this book but can be readily found in Parrat [10]. In addition, at slightly larger angles, satellite peaks are observed, from which using (7.9.2), we can also obtain the period of the multilayers. Again, for the same multilayer a low angle scan (reflectivity) is shown in Figure 7.9.12b. Two satellites are observed at θ ∼2.5◦ and θ ∼4.9◦ , consistent with a period of 18.4 Å.

OUP CORRECTED – FINAL

Methods of X-Ray Diffraction 449 Pt(111) Ag(111) n = –1

30 repeats

200 A Ag

GaAs (111)

n = –2

Co(111)

n=1 ×10

Intensity (a.u.)

Intensity (a.u.)

(a)

(b) N=1 N=2

1 30

35 40 Detector angle, 2θ

45

2 3 4 Incident angle, θ

5

Figure 7.9.12 X-ray diffraction from a (Co3Å Pt15Å )30 multilayer  grown epi ˚ taxially on GaAs(111) with a Ag200Å buffer layer, using Cu Kα λ = 1.54 A radiation. (a) High-angle scan with the multilayer shown in the inset. (b) A lowangle scan showing the very fine oscillations corresponding to the total thickness (∼540+200 Å) of the multilayer, and two satellites (N = 1, 2) determined by the multilayer period (∼18 Å). Adapted from [7].

7.9.5 Practical Considerations: Collimators and Monochromators For X-ray diffraction, ideally, we require photons that correspond to a single wavelength. Moreover, all the diffraction methods discussed so far in this section require a narrow X-ray beam to be incident on the specimen. In the laboratory, X-ray tubes (§2.4.1) generate characteristic X-ray lines (often, more than one) superimposed on a Bremsstrahlung background (Fig. 2.4.1). Appropriate absorption filters, as discussed in §2.4.2, are used to select a narrow wavelength out of the continuous X-ray spectrum produced by the X-ray tube (Fig. 2.4.4). In addition, the radiation from an X-ray tube is emitted in all directions; hence, methods to generate a narrow parallel beam, or collimation, in the direction of the specimen is required. Alternatively, synchrotron sources (§2.4.3) produce high-intensity X-rays that are spatially very confined and pencil like (Fig. 2.4.5), but here too the wavelength of the radiation is over a broad continuous spectrum of energy. In this case, devices to select a narrow X-ray wavelength out of the synchrotron radiation, such as monochromators, are required. We now provide a brief description of such collimators and monochromators. 7.9.5.1 Collimators A collimator is typically used to produce a narrow, parallel, cylindrical beam of X-rays. In practice, a simple pinhole design (Fig. 7.9.13) consists of a narrow cylinder with two apertures (A1 and A2 ) to define the beam, and a third aperture (A3 )—the guard aperture, which does not define the beam but stops all radiation

OUP CORRECTED – FINAL

450

X-Ray Diffraction A1

A2 A3 d

Source

δ

X-rays

l Figure 7.9.13 A pinhole collimator. The maximum angle of divergence, δ, depends on the diameter, d, of the defining aperture (A2 ), and the separation, l, between the two apertures A1 and A2 . scattered by the second defining aperture (A2 ). The apertures are normally circular in shape, but square and rectangular shapes can also be used. When cylindrical pinhole apertures are used in the laboratory, they are combined with either filters or monochromators to obtain a narrow wavelength of X-rays. For the collimator shown in Figure 7.9.13, the divergence angle of the beam, δ, can be calculated, assuming small δ, from the dimensions of the cylindrical pinhole as tan

d/2 d δ δ δ = = = sin = 2 l/2 l 2 2

(7.9.4)

Then, δ = 2d/l, where the divergence angle is in radians. The dimensions of a typical pinhole collimator are d = 0.3−0.5 mm, and l = 40−50 mm; thus, taking average values we get δ = 0.8/45 = 0.018 radians = 1.02◦ . 7.9.5.2 Monochromators We have seen earlier (Fig. 2.4.4) that a filter can be used to select a wavelength interval, out of the spectrum produced by an X-ray tube. The filter absorbs a range of unwanted radiation but allows the X-rays of interest for the diffraction experiment to go through. Alternatively, as we have seen in the design of wavelength dispersive spectrometers (§2.5.1.1), specific single crystals (Table 2.5.1) can be used to select X-rays of a specific wavelength. Applying the same principle of single crystal diffraction, we can also produce a narrow beam of X-rays within a very narrow range of wavelengths. Such devices, known as single crystal monochromators, are based on the simple application of Bragg’s law, (4.3.3). When radiation of different wavelengths is incident on a single crystal, diffracted beams are produced in different directions depending on the diffraction angle (β) and the wavelength (λ) of the X-rays. In other words, a specific wavelength out of the X-ray spectrum can be selected by choosing a specific diffraction angle (β). In the simplest monochromator, a single crystal is cut in such a way that a major set of crystallographic planes parallel to one of the crystal faces satisfies the Bragg diffraction condition. To be truly effective the interplanar spacing of the monochromator, in the specific crystal orientation, should be in the range to produce a reasonable X-ray beam at the required wavelength. Moreover, to minimize the loss of intensity, a reflection that has a small scattering angle is chosen to minimize the contributions from the Lorentz polarization factor (Fig. 7.10.4).

OUP CORRECTED – FINAL

Factors Influencing X-Ray Diffraction Intensities 451

S

D Divergence Slit

180-2θ

2θ 180-

Curved Diffracting Crystal

A particularly clever design of a very efficient X-ray monochromator uses a curved diffracting crystal (Fig. 7.9.14). The polychromatic source produces characteristic lines (Kα1 , Kα2 , Kβ, etc.) above the continuum background (Fig. 2.4.1). A curved and shaped crystal is used that has a radius twice that of the focusing circle, but shaped such that the inner surface fits the focusing circle, as shown. The position of the source, S, the angle (180–2θ)◦ , and the interplanar spacing, dhkl , of the crystal are chosen to select a specific line, such as Kα1 , and bring the divergent beam from S, to a focus at the point, D, on the focusing circle. Further, the X-rays emerging from D are highly monochromatic and effectively serve as the source producing a narrow beam of X-rays.

7.10 Factors Influencing X-Ray Diffraction Intensities In §7.4 we derived an expression for the structure factor, Fhkl , and the intensity, Ihkl , of a specific reflection, hkl, in X-ray diffraction. In practice, this ideal intensity is modified by a variety of experimental factors. We now discuss each of the four principal factors.

7.10.1 Temperature Factor Earlier, while calculating structure factors (§7.4 and §7.5), we assumed that the crystal is a collection of atoms occupying fixed positions in the unit cells. However, the atoms vibrate about their mean or equilibrium positions with the amplitude

Figure 7.9.14 A curved crystal diffracting monochromator. The source may be a point or a line, emitted by an X-ray tube; often it is polychromatic. The diffracting planes (hkl) of the curved crystal make an angle, θ , with respect to the incident and diffracted beams. The point D is now the focused monochromatic source for X-rays of a specific wavelength (e.g. Kα1 ). See Figure 2.5.1.

OUP CORRECTED – FINAL

452

X-Ray Diffraction of the vibration, typically of the order of one-tenth the lattice spacing (§5.3.5.1), increasing as the temperature increases. As discussed in §3, the atoms in a crystal are bound to each other by various types of forces, and their equilibrium positions correspond to the energy minima. Hence, as their positions are perturbed, the atoms will return to their original positions in the form of oscillations. The collective vibration modes of all the atoms in a crystal are known as phonons. However, it is still reasonable to assume, as a first approximation, that the thermal motions of atoms in a crystal are independent of each other. In any case, these oscillations will effectively modify the electron density of each atom and alter their ability to scatter X-rays (or, for that matter, also electron and neutron probes). A larger vibration amplitude implies a larger spatial extent of the electrons, or alternatively, for the same number of electrons, an effectively lower electron density. This reduces the atomic scattering factor, fT , at any temperature, T, as well as the diffraction intensity, Ihkl (T), for a specific reflection. Typically, the period of the atomic vibration is much smaller than the time scale of an X-ray scattering experiment. As a result, only the timeaveraged displacement of the atoms with respect to their equilibrium position is required to describe the effect of thermal motion on diffraction. Without going into the specific details (see Giacovazzo, 2002), we can simply state that to accommodate the effect of temperature, the atomic scattering factor is multiplied by an attenuation or damping factor:  fT

sin θ λ



 = f0

sin θ λ

 e

 2 −B(T) sin θ

λ

= fT (s) = f0 (s)e−B(T)s = f0 (s)e−M 2

(7.10.1)

9 Peter J. W. Debye (1884–1966) was an American physical chemist who received the 1936 Nobel prize in Chemistry for his “contributions to the study of molecular structure.” 10 Ivar Waller (1888–1991) was a Swedish physicist.

Here, the subscript, 0, refers to the ideal value at 0 K. Further, the exponential factor, B(T ), at any temperature, T, is proportional to the mean square displacement,

2 ri , of the atom, i, in a direction normal to the reflecting planes. The exact value

 of ri2 is rather difficult to determine, but the dependence of B(T ) on temperature was first studied by Debye,9 who also developed an analytical expression for fT for atomic elements. The theory of X-ray scattering by lattice vibrations was further developed by Waller10 in 1925, who also provided the definitive treatment of the subject. Thus, the factor, B(T ), are now known as the Debye–Waller factors. They are listed for some representative elements in Table 7.10.1, and a complete listing of them for a wider range of elements can be found in Peng et al. [11]. In practice, the Debye–Waller factors, B(T ), are not known accurately for most crystals; thus, experimental measurements are preferred, but if they are not available, elemental values are used as a first step. At room temperature they range from ∼10 Å2 (the first column elements; Li, Na, K . . . ), to ∼1–3 Å2 (the second column elements; Be, Mg, . . . ), to ∼0.3–0.7 Å2 (for the remaining elements) in the periodic table. Needless to say, B(T ) depends on temperature, as summarized for select elements in Table 7.10.1. Finally, the intensity of a diffracted beam is now modified by the temperature factor, e−2M , with respect to the value at 0 K,

OUP CORRECTED – FINAL

Factors Influencing X-Ray Diffraction Intensities 453 1.0

e–2M

0.8 0.6 0.4

fcc

0.2

bcc

0

0

0.2

0.4 sin θ λ

Figure 7.10.1 Variation of the temperature factor for FCC and BCC iron at 280◦ K as a function of the scattering parameter.

0.6

Table 7.10.1 Debye–Waller factors, B(T ), in Å2 , for some representative elements as a function of temperature. A more detailed list is in [12], from which this is taken. T (K)

C (Dia)

Na (BCC)

Fe (BCC)

Fe (FCC)

Zn (HCP)

Au (FCC)

90 120 170 220 270 280

0.1300 0.1311 0.1336 0.1370 0.1412 0.1422

2.0516 2.7348 3.8721 5.007 6.1389 6.3648

0.1493 0.1332 0.1886 0.2441 0.2995 0.3106

0.1715 0.2286 0.3238 0.4189 0.5140 0.5330

0.3512 0.4683 0.6632 0.8580 1.0526 1.0915

0.1908 0.2544 0.3602 0.4659 0.5714 0.5925

and e−2M decreases with scattering parameter, example for Fe, both FCC and BCC, at 280◦ K.



sin θ

λ

 ; Figure 7.10.1 shows an

7.10.2 Absorption or Transmission Factor As X-rays travel through a crystal, both the incident and diffracted beams are absorbed. The absorption reduces the intensity of the X-rays traveling through the material by a factor proportional to the path-length, l, travelled and the linear absorption coefficient, μ, of the material: I = I0 e−μl

(2.4.2)

where I 0 is the incident and I the diffracted intensities. Strictly speaking, the pathlength is dependent on the precise location of the scattering event (Fig. 7.10.2), as well as the incident and scattering angles. Normally, we are interested in the scattered or transmitted intensity, which is defined by the transmission factor, T (θ ) = I /I0 , and to calculate it for the entire crystals of volume, V, we integrate (2.4.2) to get

A B C Figure 7.10.2 X-rays incident on three different points in the crystal with different primary and diffracted path lengths.

OUP CORRECTED – FINAL

454

X-Ray Diffraction T (θ ) =

1 V



e−μ(p+d) dv

(7.10.2)

v

where p and d, are the path lengths of the primary and diffracted beams, respectively, travelled in the crystal after being scattered at any volume element, dv. Recall (§2.4.2) that the mass absorption coefficients, μm i = μi /ρi , are known (Table 2.4.1) for most elements, i, and it is easy to calculate the linear absorption coefficient, μ, for any material as μ=ρ



wi μm i

(7.10.3)

i

where ρ is its overall density, and wi are the mass fractions of its constituent elements. Note that μm i is smaller for lower atomic numbers and for shorter wavelengths. As a result, the absorption correction becomes more important for heavier elements and longer wavelengths. An analytical evaluation of T (θ ), (7.10.2), would depend on the specific beam path, which is a function of the reflection, hkl, and the overall shape of the crystal. Unfortunately, the integral, (7.10.2), cannot be calculated analytically even for the simplest crystal shapes, and hence, numerical methods are used [12]. Even though these results are beyond the scope of the book, we summarize the key results and trends. First, the transmission coefficient, T(θ), depends on the geometry of the diffraction method; as such, we only consider the two most common methods of diffractometry and the Debye–Scherrer camera. Second, the absorption is always greater for the low-θ reflections, and the difference between low-θ and high-θ reflections decreases as the linear absorption coefficient decreases. In diffractometry, for a beam of fixed cross-section, when θ is small the area of illumination on the specimen surface is large, but its depth of penetration is small; alternatively, when θ is large, the area of illumination is small, but the depth of penetration is large. As a result, the volume of interaction between the X-ray beam and the specimen remains constant and independent of θ. Thus, T(θ) is independent of θ, and is a constant for a material: Tdiffractometry =

1 2μ

(7.10.4)

Thus, for all reflections, hkl, the absorption reduces the intensity by the same factor, and if ratios of intensities are being considered, the absorption effect cancels out and it can be ignored. In the Debye–Scherrer camera, a thin needle-like cylindrical specimen is used, and for this geometry T(θ) is difficult to calculate. However, the functional form of the dependence of absorption with angle, θ , mentioned earlier (i.e. the absorption is always greater for the low-θ reflections), is exactly opposite of the temperature factor (Fig. 7.10.1). Hence, it is reasonable to assume that the two effects of

OUP CORRECTED – FINAL

Factors Influencing X-Ray Diffraction Intensities 455 absorption and temperature cancel out when the Debye–Scherrer camera is used. Further details, specifically on absorption effects in diffraction, can be found in Giacovazzo (2002) and Cullity (1978).

7.10.3 Lorentz Polarization Factor We have already seen (7.2.8) that when a totally unpolarized beam of X-rays is scattered in different directions, even by a single electron, its intensity is affected by the polarization factor, P(θ), given by P (θ ) =

 1 1 + cos2 2θ 2

(7.10.5)

where θ is the Bragg angle. Note that, in theory, 0.5 < P (θ) < 1.0, but in practice, the variation is less substantial. Moreover, there are three other geometrical corrections that apply, which are combined with the polarization and together written as the Lorentz polarization factor, LP (θ ). In §7.8 we saw that diffraction takes place not only at the exact Bragg orientation, but also when the orientation deviates from the exact angle. In the reciprocal formulation, or the Ewald sphere construction, diffraction is considered to have occurred when the Ewald sphere crosses a reciprocal lattice node, which, as mentioned (§7.8), has a finite volume in reciprocal space. Now, as the crystal is rotated (Fig. 7.10.3) at constant angular velocity, ω, if a specific reciprocal lattice node remains in diffracting position for a longer time, then its intensity can be higher. Depending on the method used, the time a node is in diffracting condition depends on the position and size of the node, as well as the velocity with which it is swept past the Ewald sphere. In the simplest analysis, the Lorentz factor, LP (θ), can be shown to be LP (θ ) =

ω vn λ

(7.10.6)

where vn is the radial component of the velocity of the reciprocal lattice node. The linear velocity, v, of the point, P, in Figure 7.10.3 is v = |g| ω

(7.10.7)

vn = |g| ω cos θ

(7.10.8)

and its radial component, vn , is

Now, when the Bragg condition is satisfied |ghkl | =

2π 2π = 2 sin θ dhkl λ

(7.10.9)

OUP CORRECTED – FINAL

456

X-Ray Diffraction

ω|g hk θ

l|

P

ghkl

ω

2θ

Figure 7.10.3 Lorentz correction, L(θ ), for the rotation of a crystal with angular velocity, ω, about an axis normal to the plane defined by the incident and diffracted X-ray beams. from which we get vn = 2π

ω (2 sin θ cos θ ) λ

(7.10.10)

and substituting in (7.10.6), we get LP (θ) ≈ sin−1 2θ

(7.10.11)

Now, for a powder crystalline specimen, we have seen that the crystal diffracts X-rays in the surface of a cone, with their tips ending on the surface of the Ewald sphere with an opening angle of the cone of 2θ (Fig. 7.9.4 and Fig. 7.9.8). However, only a small fraction of the total intensity scattered by a family of lattice planes, {hkl}, is intercepted by the detector. For example, in a Debye– Scherrer camera of radius, R, the radius of the diffracted cone, rhkl , at the point of intersection with the camera is R sin 2θ, with a total length of the diffraction line given by 2π rhkl = 2π R sin 2θ. Thus, the diffraction intensity per unit length of the diffraction line is proportional to sin−1 2θ , which is the second trigonometric correction. Lastly, for a powder specimen with randomly oriented grains, the number of grains oriented favorably for diffraction with an angle 2θ is proportional to cos θ . Combining all these three factors, we get the Lorentz-polarization factor     1 + cos2 2θ LP (θ ) ≈ 1 + cos2 2θ sin−1 2θ sin−1 2θ cos θ ≈ 2sin2 θ cos θ

(7.10.12)

where the proportionality constant has been dropped and only the angular dependence (Fig. 7.10.4) is included.

OUP CORRECTED – FINAL

Factors Influencing X-Ray Diffraction Intensities 457

LP (θ)

30 20 10 0

0

20

40 60 Bragg angle, θ

Figure 7.10.4 The variation of the Lorentz-polarization factor with the Bragg angle.

80

7.10.4 Multiplicity This last correction factor takes into account the total number of variants of a specific family of lattice planes {hkl} that contribute to a given diffracted intensity, especially in a polycrystalline specimen with randomly oriented crystallites. The total number of planes in a given family, known as the multiplicity, Phkl , depends on the specific crystal structure. For example, consider the {hkl} planes, where cubic = 48, P hexagonal = 24, P tetragonal = 16, P orthorhombic = 8, h = k  = l. Then, Phkl hkl hkl hkl monoclinic = 4, and P triclinic = 2. Thus, the intensity, I , must be multiplied by Phkl hkl hkl crystal Phkl to give the observed intensity.

7.10.5 Corrected Intensities for Diffractometry and the Debye–Scherrer Camera crystal

We can now include all these corrections factors: multiplicity, Phkl , Lorentzpolarization, LP (θ ), transmission, T (θ ), and Debye–Waller, B(T ), and write a general expression for the observed diffraction intensity as I = Ihkl Phkl LP (θ ) T (θ ) e−2M = |Fhkl |2 Phkl cryst

cryst

1 + cos2 2θ 2 sin2 θ cos θ

T (θ ) e−2M (7.10.13)

which simplifies for a polycrystalline specimen as cryst

Idiffractometry = |Fhkl |2 Phkl

1 + cos2 2θ

1 −2M e 2 sin θ cos θ 2μ 2

(7.10.14)

for diffractometry, and cryst

ID−S = |Fhkl |2 Phkl for Debye–Scherrer patterns.

1 + cos2 2θ 2 sin2 θ cos θ

(7.10.15)

OUP CORRECTED – FINAL

458

X-Ray Diffraction Example 7.10.1: Powder diffraction data, using Mo Kα X-rays   ˚ , for two different samples are summarized in the table λ = 0.71 A below. Assuming that the samples are cubic (FCC, BCC, diamond, or simple), index the observed reflections/peaks for both specimens. Find the lattice parameters and identify the structures of the two samples. How unique is your structural analysis?

Peak #

Sample A 2θ (Deg) I/I max

Sample B 2θ (Deg) I/I max

1 2 3 4 5 6 7 8

19.6 22.7 32.3 38.1 39.8 46.3 50.8 52.2

20.2 28.7 35.4 41.1 46.2 50.9

100 44 25 28 7 5 26 31

100 15 27 9 14 3.5

Solution: We calculate the d-spacing for each peak. We then divide the dspacing for each peak by that of the first peak, and then square it. For a  2 cubic crystal, this gives us the ratio of the squares of the indices, i.e. dd1n = h2n +k2n +ln2 . h21 +k21 +l12

The results are shown in the table, which includes the consistent

indexing and the related lattice parameter. Peak #

2θ (Deg)

d hkl (Å)

1 2 3 4 5 6 7 8

19.6 22.7 32.3 38.1 39.8 46.3 50.8 52.2

2.1 1.81 1.28 1.1 1.04 0.91 0.83 0.81

(d 1 /d n )2

hkl

a (Å)

1 1.34 (∼4/3) 2.69 (∼8/3) 3.64 (∼11/3) 4.08 (∼12/3) 5.33 (∼16/3) 6.4 (∼19/3) 6.7 (∼20/3)

111 200 220 311 222 400 331 420

3.64 3.62 3.62 3.65 3.60 3.64 3.62 3.62

Since all the indices are unmixed, this must be an FCC structure. We can develop a similar table for Sample B:

OUP CORRECTED – FINAL

Applications of X-Ray Diffraction 459

Simple Cubic

BCC

Peak #

2θ (Deg)

d hkl (Å) (d 1 /d n )2

hkl

a (Å)

hkl

a (Å)

1 2 3 4 5 6

20.2 28.7 35.4 41.1 46.2 50.9

2.02 1.43 1.17 1.01 0.90 0.83

100 110 111 200 210 211

2.02 2.02 2.03 2.02 2.01 2.03

110 200 211 220 310 222

2.86 2.86 2.87 2.86 2.85 2.88

1 2.0 3.0 4.0 5.0 ∼6.0

It is possible to index the diffraction data both as a simple cubic with ˚ and a BCC structure with a = 2.86A. ˚ However, we also have the a = 2.02 A intensities, which we can analyze as follows: Peak #

1 2 3 4 5 6

2θ (Deg)

20.2 28.7 35.4 41.1 46.2 50.9

Lorentz polarization factor (LPF)

Simple Cubic

BCC

hkl

Multiplicity, m

I LPF×m

hkl

Multiplicity, m

I LPF×m

31.06 14.87 9.45 6.8 5.22 4.2

100 110 111 200 210 211

6 12 8 6 24 24

0.54 0.08 0.34 0.25 0.13 0.09

110 200 211 220 310 222

12 6 24 12 24 8

0.27 0.17 0.12 0.11 0.11 0.10

We calculate the Lorentz polarization factor (7.10.12), and the multiplicity based on the crystal structure for each reflection. We then divide the observed intensity by the product of the Lorentz polarization factor and the multiplicity. If the fit is reasonable, the resultant value should be only proportional to the atomic scattering factor squared, and should decrease monotonically with the scattering angle. Thus, we can infer that Sample B is BCC (in this case, it happens to be FeBCC ).

7.11 Applications of X-Ray Diffraction X-ray diffraction is a versatile tool, used in materials characterization and analysis in a wide range of disciplines. We conclude this chapter with some typical applications of X-ray diffraction, both here on Earth and on Mars [13]!

OUP CORRECTED – FINAL

460

X-Ray Diffraction

7.11.1 Measurement of Lattice Parameters How do we measure the lattice spacing, dhkl , of a crystalline material most accurately by X-ray diffraction? As a first step, we define the resolving power, ∂d/d, of a diffraction experiment by taking the derivative of the Bragg equation, λ = 2d sin θ , for a fixed wavelength, λ, used in the experiment as 0 = 2 sin θ∂d + 2d cos θ ∂θ

(7.11.1)

which gives the resolving power cos θ ∂d =− ∂θ = − cot θ∂θ d sin θ

(7.11.2)

Thus, for a fixed value of 2 ∂θ , which defines the minimum separation between reflections that can be measured in a specific instrument, we can get the smallest resolving power when cot θ has the smallest value possible. We know from the table of values of cot θ that it has largest values for θ ∼0◦ , and then rapidly decreases as the Bragg angle, θ → 90◦ . Thus, where possible, reflections with large values of 2θ should be used for the most accurate measurements of lattice parameters. In diffractometry, physical limitations of bringing the detector close to the X-ray source restricts 2θ to be less than 150◦ . However, large 2θ angles, especially in the back-reflection geometry, are routinely observed in Debye–Scherrer cameras, which is the principal factor driving their continued use today.

7.11.2 Crystallite or Grain Size and Lattice Strain Measurements As shown earlier (Fig. 7.8.1), the broadening of the peak, β, can be related to the crystallite size by the Scherrer equation (7.8.6). However, the broadening observed in an experiment, β expt , is also influenced by parameters such as the detector slit width, area of the specimen that is illuminated, whether the incident X-ray beam includes the kα2 line or not, etc. To get around these instrumental factors, a similar X-ray diffraction pattern is measured using a large-grain material or a single crystalline material, whose contribution to the broadening can be ignored, and obtain only the broadening, β inst , contribution from the instrument. The difference, β = βexpt − βinst , can then give the true broadening from the crystallite or grain size through the Scherrer equation, (7.8.5). In practice, the accuracy of such measurements is affected by defects, such as small angle grain boundaries, and/or lattice strain in the material. Strain in a material can be either on the macroscale or on the microscale. In the former, the whole specimen is subject to some tension or compression, and as a result the lattice spacing, dhkl , increase or decrease in the direction of strain. In X-ray diffraction, the peaks shift in position, and by measuring them for the

OUP CORRECTED – FINAL

Applications of X-Ray Diffraction 461 specimen in different positions or orientations, the magnitude and direction of the strain can be determined. In contrast, if the strain is on the microscale, the magnitude and direction of the strain vary from crystallite/grain to crystallite/grain. Then, rather than observing a peak shift, we observe a broadening of the peaks in X-ray diffraction. To describe this mathematically, we use (7.11.2), but replace ∂d d by the elastic strain, ε. Thus, ε = − cot θ ∂θ

(7.11.3)

Hence, the broadening, β strain , from the strain, defined as the full width at halfmaximum of the peak, 2∂θ, is βstrain = −

2ε = −2ε tan θ cot θ

(7.11.4)

In other words, the broadening due to crystal size (Scherrer equation) varies as sec θ, whereas the broadening due to microstrain varies as tan θ . If the goal is to experimentally determine the microstrain, the instrument contribution to the broadening should be measured first and subtracted as described. Finally, if the Young modulus for the material is known, the measured microstrain, ε, can be interpreted in terms of the stress in the material.

7.11.3 Phase Identification and Structure Refinement Powder diffraction is most commonly used for the quantitative and qualitative analysis of crystalline materials. From our discussions of powder diffraction so far, it is clear that the lattice spacing, dhkl , and the relative intensities of the diffraction peaks, Ihkl , for a polycrystalline material is dependent on the nature of the material and its specific crystalline form. As such, the analysis of powder diffraction patterns is a widely used method for the identification of unknown materials. For this purpose, starting in the 1930s, specific files containing details of the diffraction patterns of a number of known materials have been compiled. Today, these data files known as the powder diffraction files (PDF), contain more than 893,400 standard entries, and the Joint Committee for Powder Diffraction Standards (JCPDS) distributes the collection of these files, in the form of a “card” for each standard material. Figure 7.11.1. shows a typical example of a PDF card for NaCl annotated for the new user. Most of the information in the card is readily interpretable and the intensities are expressed with respect to the strongest line/peak, which is assigned an arbitrary value of 100. It is possible to match the information provided in such a “standard” card with the diffraction pattern of an unknown; for example, the card (Fig. 7.11.1) can be readily matched with the powder diffraction pattern for NaCl (Fig. 7.9.5). However, for simple materials such as NaCl even though the

OUP CORRECTED – FINAL

462

X-Ray Diffraction

1

2

3

Quality of data

4

5-628 d I/I1

5

6 7

8

2.82 100

1.99 55

1.63

3.26

15

13

Na Cl Sodium Chloride

Dia. Rad. CuKα1 λ =1.5405 Filter Ni I/I cor. I/I1 Diffractometer Cut off Ref. Swanson and Fuyat, NBS Circular 539, Vol. 2, 41 (1953) Sys. Cubic a0 5.6402 b0 α β Ref. Ibid. eα 2v D Ref. Ibid.

c0 γ nωβ 1.542

S.G. Fm3m (225) A C Z 4 Dx 2.164 eγ

mp

Sign Color colorless

An ACS reagent grade sample recrystallized twice from hydrochloric acid X–ray pattern at 26°C Merck Index, 8th Ed., P. 956 Halite - galena - periclase group.

I/I1 dA 13 3.258 2.821 100 55 1.994 2 1.701 15 1.628 6 1.410 1 1.294 11 1.261 7 1.1515 1 1.0855 2 0.9969 1 0.9533 3 0.9401 4 0.8917 1 0.8601 3 0.8503 2 0.8141

hkl 111 200 220 311 222 400 331 420 422 511 440 531 600 620 533 622 444

(Halite) I/I1

dA

hkl

9 Figure 7.11.1 Annotated PDF card for NaCl: (1) File number. (2) Three strongest lines, corresponding to the 200, 220, and 222 reflections. (3) Lowest-angle line. (4) Chemical formula and name. (5) Description of diffraction method used. (6) Crystallographic data. (7) Optical and other data. (8) Description of specimen. (9) Details of the diffraction pattern. Reproduced from the JCPDS powder diffraction data file. See Figure 7.9.5 for the θ -2θ XRD scan of NaCl, which matches very well with this PDF file. matching can be done manually by inspection, for more complex materials/phases, these days the matching is done largely using computer methods. Finally, if there are more than one crystallographic component or phases in the specimen, they can be simultaneously identified. However, a quantitative analysis is more complicated because the relationship between the intensities and the amount of a given phase in a mixture is nonlinear due to absorption effects. Details of such analysis of phase mixtures is beyond the scope of this book, but can be found in more advanced texts such as Klug and Alexander (1974). For polycrystalline materials with large unit cell volumes and/or low symmetry, indexing of diffraction patterns is rather difficult because of the frequently overlapping peak intensities. However, if a rudimentary or imperfect structural model of the material is available, the observed diffraction intensity can be compared with

OUP CORRECTED – FINAL

Applications of X-Ray Diffraction 463 that calculated based on the model. Such a least-squares refinement, originally pioneered by Rietveld [13], is now routinely used to refine structural details of powder specimens, with results comparable to that of single crystals. Example 7.11.1: Show that the observed X-ray powder diffraction (Fig. 7.9.5) is that of NaCl. Assume Cu Kα (1.54Å) radiation was used. Solution: We solve this as a table: Peak position (2θ ) in Figure 7.9.5. Peak intensity Relative intensity λ d = 2 sin θ

27.5 31

46

54

56

67

73

650 6100 3000 200 900 400 100 10.7 100 49.1 3.3 14.75 6.6 1.6 3.24 2.88

1.97

1.7

1.64

1.4

75

84

900 400 14.75 6.6

1.29 1.265 1.15

Both the peak positions (d-spacings) and the relative intensities match very well with the PDF card for NaCl (Figure 7.11.1). In fact, the indexing of all the observed peaks as show in Figure 7.9.5 is indeed correct.

In the Rietveld method, the entire profile of the powder pattern is fitted to a calculated pattern consisting of all the Bragg reflections, assuming each of them to be of a specific functional form (typically a Lorentzian or a Gaussian function, or a sum of such functions is used) and centered at their respective Bragg reflection position. These complex shapes for the peaks are required to accommodate the actual shapes of the X-ray emission lines when laboratory sources are used; however, if synchrotron sources are used a simple Gaussian shape may suffice. The fitting program minimizes the residual function S=



wi {Yio − Yic }2

(7.11.5)

i

where wi is the least-squared weight, given by 2 2 2 w−1 i = σi = σip + σib

(7.11.6)

and σip is the standard deviation associated with the peak, and σip is that associated with the background intensity, Yib . Yio is the intensity observed at the i th step of 2θi , and Yic is the calculated intensity contributions from the Bragg reflection and background Yic = k

 hkl

 mhkl Lhkl |Fhkl |2 G θihkl + Yib

(7.11.7)

OUP CORRECTED – FINAL

464

X-Ray Diffraction where k is a scale factor, m is the multiplicity factor, F is the structure factor, θi = 2θi − 2θc , where 2θc is the calculated position for the Bragg reflection, and G is the profile function, all for the reflection hkl. Various parameters can be adjusted for refinement and these include details of the unit cell, atomic positions, the Debye–Waller thermal parameters and the parameters that define the peak– shape function, G. Figure 7.11.6 shows an example of Rietveld analysis.

7.11.4 Chemical Order–Disorder Transitions Consider a simple binary alloy composed of elements A and B. At high temperatures, atoms of A and B are distributed randomly in the lattice, occupying each atomic site with the same probability. Such a solid solution is called a disordered alloy. Now, when the temperature is lowered, the atoms arrange themselves such that the A atoms occupy a certain lattice site in an orderly manner, and the B atoms occupy a different set of lattice sites. Thus, we have two sublattices in the crystal, one with atom A, and the other with atom B. Such a solid solution is known as chemically ordered, or a superlattice. Note that the superlattice must conform to one of the 14 Bravais lattices (§4.1.2); as such, the ordering is only a change in the lattice type, and calling it a superlattice, though common, is not strictly correct. In alloys, when this type of chemical order of atoms is observed over large distances compared to the unit cell, it is known as long-range order. Typically, an alloy disordered at high temperatures transitions into an ordered one as the temperature is lowered at a characteristic temperature, T c , called the ordering temperature. Alloy systems that show such changes are said to exhibit order– disorder transitions. Classic examples of ordered alloys in the cubic system (Fig. 7.11.2) are referred to as the CuZn (BCC)-, CuAu (FCC)-, and Cu3 Au(FCC)type ordered structures. It is reasonable to expect alloys that exhibit order–disorder transitions to exist in some intermediate state, where the degree of order is incomplete. The degree of order can be quantitatively described by an order parameter, S, defined as S=

rA − XA 1 − XA

(7.11.8)

where rA is the fraction of A lattice sites occupied by the A atoms, and XA is the fraction of A atoms in the alloy. For a completely random distribution of atoms in a disordered alloy, rA = XA , and S = 0, and for a perfect long-range ordered alloy, rA = 1, and S = 1. The presence of long-range order, and specifically, the order parameter, can be measured by X-ray, electron, or neutron diffraction. In the case of alloys, when there is negligible change in the unit cell size upon ordering, there will be no observable change in the position of the X-ray diffraction lines; however, ordering significantly affects the peak intensities. Additional reflections, referred to as superlattice lines (see §7.5.4), are often observed signaling a change in the

OUP CORRECTED – FINAL

Applications of X-Ray Diffraction 465 (a)

(b)

(c)

Zn

Au

Au

Cu

Cu

Cu

Figure 7.11.2 Unit cells of ordered binary alloys with atom positions as indicated. (a) The CuZn (a Cubic P lattice, which transforms from the Cubic I lattice in the disordered phase). Note that this structure is similar to the CsCl structure discussed in §7.5.4 (b) The CuAu structure (a Cubic P lattice, but strictly speaking there is a tetragonal distortion in the ordered phase, and as such, it is a Tetragonal P lattice. The corresponding disordered phase is the Cubic F lattice). (c) The Cu3 Au structure (a Cubic P lattice, with a basis of 3Cu+Au). In the disordered phase, it is a Cubic F lattice. Bravais lattice upon ordering. In addition, other types of diffraction lines, known as fundamental reflections, are observed in the same position, for both the ordered and disordered states of the alloy. Note that, in general, the superlattice reflections have a lower intensity compared to the fundamental. Nevertheless, the order parameter, S, can be obtained from experimentally measured diffraction data by a simple ratio of observed intensities:  Ff S= Fs

mf (LP )f Is ms (LP )s If

(7.11.9)

where the subscripts s and f refer to the superlattice and fundamental lines, m is the multiplicity, (LP ) is the Lorentz polarization factor (7.10.12), F is the structure factor, and I is the observed intensity, often measured as the integrated area of the peak. For the CuZn alloy (Fig. 7.11.2a) by calculations similar to those in §7.5.2 for a BCC crystal, we can show that the structure factors are FCuZn = fCu + fZn

for (h + k + l) even

FCuZn = fCu − fZn

for (h + k + l) odd

(7.11.10)

Thus, the fundamental lines, (h + k + l) even, are observed in both ordered and disordered alloys. However, the superlattice lines, (h + k + l) odd, with intensities proportional to the degree of order, S, are only observed for the ordered alloys. As a first approximation, assuming all else is equal, we write the ratio of observed intensities, based on their respective structure factors, as

OUP CORRECTED – FINAL

X-Ray Diffraction 400 111 100 110 303 K 773 K 873 K 973 K 1073 K 1173 K 1273 K 1283 K 1293 K 1303 K 20 22 24 26 28 30 32 34 36 38 40 42 2θ (degree)

Integrated intensity (arb. units)

(a)

Intensity

466

(b)

350 300 250 200 150 100 50 0 800

900

1000

1100

1200

1300

Temperature (K)

Figure 7.11.3 (a) Temperature dependence, from 300K to 1303K, of the X-ray diffraction from a specimen of (Cu0.5 Fe0.5 ) Pt3 . The fundamental (111) reflection and the superlattice (100) and (110) reflections are shown. (b) A plot of the integrated intensity of the superlattice (100) reflection as a function of temperature, showing the order-disorder transition occurs at ∼1300◦ K. Adapted from [13].

|Fs |2 |fCu − fZn |2 |29 − 30|2 Is =  2 = = ≈ 0.003% 2 If Ff  |29 + 30|2 |fCu + fZn |

(7.11.11)

As a result, the superlattice reflections in CuZn alloys are much too weak, compared to the fundamental ones, to be easily detected by X-ray diffraction. This is representative of a general problem with X-ray diffraction as the atomic scattering factors are proportional to the atomic number, and hence, the positions of atoms with similar atomic numbers in the unit cell are hard to determine. This problem is often overcome using neutron diffraction, where the scattering amplitudes are determined by the nucleus (and not the electrons) and the scattering factors vary irregularly with atomic number (see §8.8). Based on this example, we can expect the ratios of intensities (in the forward 502 direction) to be IIfs = 166 2 ≈ 10%, and this ordering can be readily detected by X-ray diffraction. Figure 7.11.3a shows the X-ray diffraction peaks, both fundamental and superlattice reflections, in an alloy of Pt3 (Cu0.5 Fe0.5 ), having the Cu3 Au ordered structure, measured as a function of temperature. The temperature dependence of the integrated intensity of the superlattice (100) reflections (Fig. 7.11.3b), indicates an order–disorder transition temperature of ∼1300K (see Problem 7.7).

OUP CORRECTED – FINAL

Applications of X-Ray Diffraction 467 Example 7.11.2: Consider the Cu3 Au alloy shown in Figure 7.11.2c. For the completely disordered alloy, the probability that an Au atom occupies an atomic site in the FCC lattice is ¼, and for occupation by a Cu atom it is ¾ (this follows from the stoichiometry). For the ordered alloy, the positions, un vn wn , of the basis atoms are 000 for Au, and ½ ½ 0, 0 ½ ½, and ½ 0 ½, for the three atoms of Cu. Calculate the structure factor and intensities for the disordered and ordered alloy if fAu = 79 and fcu = 29 (see also Fig. 7.11.3). Solution: The structure factor for the disordered alloy (see §7.5.1), is given by Fhkl = 4 × 1/4 × fAu + 4 × 3/4 × fCu = fAu + 3fCu

for h, k, l all even or all odd

Fhkl = 0

for h, k, l with mixed indices.

For the ordered alloy the structure factor is given by (7.5.1), but by taking the specifics of chemical ordering into consideration, i.e.        2πi 2h + 2k 2πi 2k + 2l 2πi 2l + 2h Fhkl = fAu e2πi.0 + fCu e +e +e

Thus Fhkl = fAu + 3fCu = 166

for h, k, l all even or all odd (fundamentals)

Fhkl = fAu − fCu = 50

for h, k, l with mixed indices (superlattices)

with corresponding intensities given by Ihkl = |Fhkl |2 = (fAu + 3fCu )2 = (79 + 3 × 29)2 = 1662 = 27556

for h, k, l all even or all odd

Ihkl = |Fhkl | = (fAu − fCu ) 2

2

= (79 − 29)2 = 502 = 2500

for h, k, l with mixed indices.

7.11.5 Short-Range Order (SRO) and Diffuse Scattering From Figure 7.11.3b, we notice that above the critical temperature, T c , the longrange order disappears, and the intensity of the superlattice reflection goes to zero. This suggests that the atomic arrangement in the alloy, for T > Tc , may be random. However, if you look carefully at the diffuse scattered intensity (Fig. 7.11.4) that forms the background of the diffraction pattern, it often shows a small peak,

OUP CORRECTED – FINAL

X-Ray Diffraction

Figure 7.11.4 The measured intensity from a single crystal of Cu3 Au, along the h00 direction for 405◦ C and 450◦ C. The short-range order peaks at 100 and 300 are normalized with respect to the fundamental 200 reflection. Adapted from [14].

Intensity (normalized)

468

405°C 450°C

10 5 0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

h

indicating that the material is not completely disordered and a correlation between nearest neighbors can persist. For alloys consisting of a matrix and guest atoms, the guest atoms can interact either with the matrix atoms or with each other. Initially, for dilute alloys, the guest atoms mainly interact with the host/matrix atoms, but with increasing concentration, guest–guest interactions also become important. This gives rise to two competing possibilities; one, where unlike atoms prefer to be nearest neighbors, a tendency known as short-range order (SRO), and typically, observed in monovalent matrices. Alternatively, like atoms attract each other, a process known as clustering, and typically, observed in multivalent matrices. Now, consider the Cu3 Au structure shown in Figure 7.11.2c. When the crystal is perfectly long-range ordered, an Au atom at 000 is surrounded by 12 Cu atoms at ½ ½ 0 (and equivalent sites, three shown in the figure). Similarly, every Cu atom is surrounded by four Au atoms and eight Cu atoms. Ideally, for T > Tc , this ordered arrangements breaks down and this atomic arrangement becomes truly random. Then, any given Au atom will have nine Cu atoms (= 3/4∗ 12, based on the structure and composition) as nearest neighbors. However, above T c it is observed that there are, on average, 11 Cu atoms around each Au atom. In fact, such SRO is a general effect and any solid solutions exhibiting long-range order also exhibits SRO above T c , with the degree of SRO decreasing with increasing temperature. Figure 7.11.4 shows X-ray diffraction intensities for Cu3 Au, taken at two different temperatures above Tc = 390◦ C. SRO maxima, i.e. 100 and 300 type reflection, are observed at the same positions as the superlattice reflections, with intensities ∼6–12 % of the fundamental 200 reflection. Notice that if a single crystal specimen is not used, other sources of diffuse scattering, primarily temperature, mask the SRO intensities. It is possible to quantify the degree of SRO using an appropriate parameter based on such measurements. Further details can be found in Schwartz and Cohen (1987).

7.11.6 In Situ X-Ray Diffraction at Synchrotrons X-ray diffraction is an outstanding method for structure analysis, with the added advantage of substantial penetration (e.g. compared to electrons; see §8),

OUP CORRECTED – FINAL

Applications of X-Ray Diffraction 469 allowing for studies of bulk materials and specimens in complicated environments. Furthermore, the high photon flux available in third-generation synchrotron sources (§2.4.3) makes it possible to collect X-ray diffraction data in relatively short time scales, and carry out in situ measurements (§1.5). Unlike the measurements of chemical order–disorder transitions described in §7.11.4, which strictly measure the thermodynamic end states, of particular importance here are measurements to understand processes such as phase transitions, crystal growth, and the role of buried interfaces in many natural processes. For example, in crystallization or thin film deposition, an atom transforms from being completely disordered in the liquid or gas phase to a crystalline form in the solid. To understand such processes as a function of thermodynamic variables, the probe should minimally interfere with the process of transformation. However, even though charged particles, particularly electrons, can be used for in situ measurements, they would have difficulty reaching buried interfaces, nor would they be able to penetrate gases under high pressure. That leaves X-rays and neutrons as potential probes. The latter lacks general availability of high flux sources (spallation sources, §5.2.3, are not common), leaving only X-ray photons at synchrotron sources as meaningful probes for such experiments. In addition, X-ray scattering satisfies the Born approximation (§8.2) and such kinematic scattering lends itself to relatively easy interpretation of diffraction data. Further practical details of implementing such in situ diffraction studies in synchrotrons are beyond the scope of this book, but interested readers are referred to the monograph edited by Ziegler et al. (2014), which gives a good introduction to such measurements, with emphasis on studies of thin film growth and deeply buried interfaces.

7.11.7 X-Ray Diffraction Measurements on Mars We end this chapter with an interesting and practical example from geology: the first X-ray diffraction measurements of the crystal structure of various minerals from soils and drilled samples found on Mars conducted by the Curiosity rover [15]. In summary, these measurements reveal an abundance of primary basaltic minerals, amorphous components, and varied hydrous alteration products, including phyllosilicates [15]. As shown in Figure 7.11.5a, the X-ray diffraction instrument, called CheMin, is a shoe box-sized apparatus, based on a variation of the Debye–Scherrer camera in the transmission geometry. It contains a micro-focus X-ray source, a pinhole for collimation, a rotating specimen holder (which holds both the collected Martian samples, as well as standards from Earth, e.g. the beryl-quartz powder used for calibration Fig. 7.11.5c), and a CCD camera for detection of the scattered X-rays. All the specimens are powders, and the ring patterns observed on the CCD camera (Fig. 7.11.5c) are radially integrated to obtain the X-ray diffraction patterns (Fig. 7.11.5b). The specimens collected on Mars are put into Mylar or Kapton specimen holders, which produce minimal background; however, the Al light shield used to protect the CCD detector introduces distinct background peaks at 2θ = 25.8◦ and 32.1◦ (Fig. 7.11.5d).

OUP CORRECTED – FINAL

470

X-Ray Diffraction Pinhole collimator X-ray beam (b) mi c X- ro-fo ray cu t ub s e

XRD

Sample holder

(a)

CCD 10

0

Intensity (counts)

(c)

20

30

50

40

9000 8000 7000 6000 5000 4000 3000 2000 1000 0

2θ

(d)

3

10

30

20

40

50

2θ

Figure 7.11.5 (a) A miniature diffractometer, CheMin, sent to Mars on the Curiosity rover, working on the transmission geometry and using an energy-discriminating CCD two-dimensional detector. (b) The two-dimensional image in (a) can be integrated to obtain the X-ray powder diffraction data. (c) A two-dimensional X-ray diffraction pattern of the beryl-quartz standard measured on Mars. (d) A comparison of the XRD patterns from empty Kapton (black) and Mylar (red) specimen holders. Circled peaks at 25.8◦ and 32.1◦ are due to scattering from the Al light shield on the CCD detector. Adapted from [15].

A typical X-ray diffraction pattern (Fig. 7.11.6) integrated from the twodimensional image (inset), recorded on the CCD camera, for Martian Rocknest is compared with the calculated plot, and refined with the Rietveld method [16]. Needless to say, the Rietveld refinement provides a very good fit to the observed data, with negligible difference or residue, except at 2θ = 25.8◦ , which corresponds to the background peak for scattering from the Al light shield for the CCD detector. Further details of the analysis of other Martian minerals, using CheMin, can be found in the original manuscript [15].

OUP CORRECTED – FINAL

Counts (a.u)

Summary 471

5

10

15

20

25 2θ

30

35

40

45

50

Figure 7.11.6 X-ray data from a specimen of the Rocknest aeolian bedform (dune) on Mars. Inset shows the two-dimensional XRD pattern. Observed (blue, integrated from the two-dimensional image in the inset) and calculated (red) plots from Rietveld refinement using data for Rocknest. The calculated background (black line) is inscribed at the base of the observed pattern and the gray pattern at the bottom is the difference plot (observed − calculated). The difference at ∼25.8◦ , circled, is due to scattering from the Al light shield on the CCD. Adapted from [15].

Summary X-rays with a wavelength on the order of the atomic spacing in solids are perfectly suited for diffraction studies from a wide range of materials. Historically, X-ray diffraction has been instrumental in our understanding of the structure, crystallography, or arrangement of atoms and molecules in many important biological, geological, or technological materials. X-rays are predominantly scattered by the electrons in solids. This can be elastic and coherent (Thomson) scattering with a well-defined phase shift of half the wavelength. Or, it can be inelastic and incoherent (Compton) scattering with the change in wavelength dependent only on the scattering angle. The net scattering of X-rays by an atom is given by the atomic scattering factor, which has the largest value corresponding to its atomic number (Z) for forward scattering (θ = 0◦ ) and decreases with increasing scattering angle; further, it also decreases with decreasing wavelength of the X-rays. Hence, the scattering factor   is parameterized as a function of the variable, s = sinλ θ , and is tabulated as the Cromer–Mann parameters for elements and ions. Diffraction intensities from a crystal result from a summation of the contributions of all the atoms in the unit cell, taking their phases also into consideration. The amplitude of the diffracted wave, thus calculated for any reflection, is known

OUP CORRECTED – FINAL

472

X-Ray Diffraction as the structure factor, Fhkl , and its square gives the intensity. If the coordinates of all the atoms in the unit cell are known, the structure factor can be calculated for any diffracted reflection, corresponding to a specific set of lattice planes (hkl), as illustrated for a variety of crystal structures in §7.5. For single crystals, the diffraction pattern reflects the underlying symmetry (point group) of the crystal. However, even if the crystal does not have a center of symmetry, the corresponding diffraction pattern will be centrosymmetric (Friedel law). Further, certain symmetry elements (glide planes and screw axes), if present in a crystal, give rise to systematic absences of reflections in diffraction. In polycrystalline materials, the diffracted beam is broadened inversely proportional to the thickness of the grain size (Scherrer equation). Ideal X-ray intensities, given by the square of the structure factor, are modified by four principal experimental factors. The temperature factor accounts for the thermal vibration of the atoms about their equilibrium position in the unit cell. Known also as the Debye–Waller factors, they are tabulated for most crystals. As X-rays travel in a crystal, both the incident and diffracted beams are absorbed in a manner proportional to the path length and the linear absorption coefficient in the material. Further, an unpolarized beam of X-rays is scattered differently in different directions, even by a single electron; this and other angular dependence are accounted for by the Lorentz-polarization factor. Finally, the total number of variants (multiplicity) of a family of lattice planes have to be included, especially in polycrystalline materials. Bragg’s law and its equivalent in reciprocal space (Laue criterion and the Ewald sphere construction) are the basis of all X-ray diffraction measurements. There are two variables, θ and λ, that can satisfy the diffraction conditions for a crystal with a given lattice spacing, dhkl . We can keep θ fixed and vary λ (Laue method) and apply it to the study of single crystals. Alternatively, we can keep λ fixed and vary θ, which has a number of variations for studies of powders (Debye–Scherrer method), both polycrystalline materials and single crystals (diffractometry), and even thin films and multilayers (reflectivity). X-ray diffraction finds a wide range of applications in the analysis and development of materials. The representative examples discussed here include measurements of lattice strain, identification of phases and structure refinement, studies of chemical order–disorder transitions, and revealing the structure and function of biological molecules, including that of DNA. .............................................................................. FURTHER READING

Cullity, B. D. Elements of X-Ray Diffraction. Boston: Addison-Wesley, 1978. Giacovazzo, C., H. L. Monaco, G. Artioli, D. Viterbo, G. Ferraris, G. Gilli, G. Zanotti, and M. Catti. Fundamentals of Crystallography. Oxford: Oxford University Press, 2002. de Graef, M., and M. McHenry. Structure of Materials. Cambridge: Cambridge University Press, 2007.

OUP CORRECTED – FINAL

References 473 Hammond, C.The Basics of Crystallography and Diffraction. Oxford: Oxford University Press, 2006. Klug, H. P., and L. E. Alexander. X-Ray Diffraction Procedures for Polycrystalline and Amorphous Materials. New York: Wiley, 1974. Preuss, E., B. Krahl-Urban, and R. Butz. Laue Atlas. New York: Wiley, 1974. Schwartz, L. B., and J. B. Cohen, Diffraction from Materials. 2nd ed. New York: Springer-Verlag, 1987. Warren, B. E.X-ray Diffraction. New York: Dover, 1990. Woolfson, M. M.An Introduction to X-Ray Crystallography. Cambridge: Cambridge University Press, 1997. Ziegler, A., H. Graafsma, X. F. Zhang, and J. W. M. Frenken, eds. In Situ Materials Characterization Across Spatial and Temporal Scales, Ch. 2. New York: Springer, 2014. .............................................................................. REFERENCES

[1] Watson, J. D., and F. H. C. Crick. “Molecular Structure of Nucleic Acids: A Structure for Deoxyribose Nucleic Acid.” Nature 171 (1953): 737–8. [2] Franklin, R. E., and R. G. Gosling. “Evidence for 2-Chain Helix in Crystalline Structure of Sodium Deoxyribonucleate.” Nature 172 (1953): 156–7. [3] Cromer, D. T., and J. B. Mann. “X-Ray Scattering Factors Computed from Numerical Hartree–Fock Wave Functions.” Acta Crystallographica Section A 24, no. 2 (1968): 321–4. [4] Wood, I. G., P. Thomson, and J. C. Mathewman. “A Crystal Structure Refinement from Laue Photographs Taken with Synchrotron Radiation.” Acta Crystallographica Section B 39, no. 5 (1983): 543–7. [5] Moffat, K., D. Szebenyi, and P. Bilderback. “X-Ray Laue Diffraction from Protein Crystals.” Science 223 (1984): 1423. [6] Moffat, K. “Time-Resolved Macromolecular Crystallography.” Annual Review of Biophysics and Biophysical Chemistry 18 (1989): 309_32. [7] Blomqvist, P., K. M. Krishnan, and D. E. McCready. “Growth of ExchangeBiased MnPd/Fe Bilayers.” Journal of Applied Physics 95, no. 12 (2004): 8019. [8] Stampe, P. A., and R. J. Kennedy. “X-ray Characterization of MgO Thin Films Grown by Laser Ablation on SrTiO3 and LaAlO3 .” Journal of Crystal Growth 191, no. 3 (1998): 478–82. [9] Cho, N.-H., K. M. Krishnan, C. A. Lucas, and R. F. C. Farrow. “Microstructure and Magnetic Anisotropy of Ultrathin Co/Pt Multilayers Grown on GaAs (111) by Molecular-Beam Epitaxy.” Journal of Applied Physics 72, (1992): 5799. [10] Parratt, L. G. “Surface Studies of Solids by Total Reflection of X-Rays.” Physical Review 95, no. 2 (1954): 359–69.

OUP CORRECTED – FINAL

474

X-Ray Diffraction [11] Peng, L. M., G. Ren, S. L. Dudarev and M. J. Whelan. “Robust Parameterization of Elastic and Absorptive Electron Atomic Scattering Factors.” Acta Crystallographica Section A 52, no. 2 (1996): 257–76. [12] Lipson, H. In International Tables for X-Ray Crystallography, Vol II: Mathematical Tables, edited by J. S. Kasper and K. Londsdale, 291–315. Birmingham: The Kynock Press, 1959. [13] Ahmad, E., M. Takahashi, K. Iwasaki, and K. Ohshima. “X-Ray Diffraction Study of Order–Disorder Phase Transition in CuMPt6 (M=3d Elements) Alloys.” Journal of the Physical Society of Japan 78, no 1 (2009): 014601. [14] Moss, S. C. “X-Ray Measurement of Short-Range Order in Cu3Au.” Journal of Applied Physics 35 (1964): 3547. [15] Bish, D., D. Blake, D. Vaniman, P. Sarrazin, T. Bristow, C. Achilles, P. Dera, S. Chipera, J. Crisp, R. T. Downs, J. Farmer, M. Gailhanou, D. Ming, J. M. Morookian, R. Morris, S. Morrison, E. Rampe, A. Treiman, and A. Yen. “The First X-Ray Diffraction Measurements on Mars.” International Union of Crystallography Journal 1, no. 6 (2014): 514–22. [16] Rietveld, H. M. “A Profile Refinement Method for Nuclear and Magnetic Structures.” Journal of Applied Crystallography 2, no. 2 (1969): 65–71. .............................................................................. EXERCISES

A. Test Your Knowledge For each statement, choose all the right answers. Note that more than one response for each question may be correct. 1. To measure lattice parameters most accurately in XRD, we prefer (a) high Bragg angles with large 2θ values. (b) low Bragg angles. (c) back reflections in a Debye–Scherrer camera. (d) to use a diffractometer over a Debye–Scherrer camera. 2. A body-centered cell will have (a) zero intensity for h + k + l = odd. (b) zero intensity for h + k + l = even. (c) the same hkl reflections with zero intensity for cubic, tetragonal, and orthorhombic unit cells. 3. An accelerating electric charge at a point, O, will generate (a) no electromagnetic radiation. (b) an electromagnetic radiation in all directions at any other point, P. (c) an electromagnetic radiation in a very specific direction at any other point, P. 4. X-rays are scattered (a) equally by all components of an atom.

OUP CORRECTED – FINAL

Exercises 475

5.

6.

7.

8.

9.

10.

11.

12.

13.

(b) primarily by the electrons of an atom. (c) only by the nucleus of the atom. The contributions of the nucleus to X-ray scattering is (a) negligible because it has a positive charge. (b) negligible because its mass is ∼103 times that of the electron. (c) significant and should always be considered. Compton scattering (a) is incoherent. (b) contributes insignificantly to the observed X-ray diffraction intensities. (c) contributes mainly to the background in X-ray diffraction. The atomic scattering factor, f, (a) decreases at large scattering angles. (b) is derived from the assumption that the charge distribution in the atom is spherically symmetric. (c) equals the atomic number for forward scattering. Diffraction intensities can be calculated using single-scattering models (a) for X-rays all the time. (b) for X-rays, if we assume that the volume of the overall crystal/ specimen is small. (c) for electron beam incidence, but generally gives erroneous results. The structure factor for an FCC crystal is (a) non-zero for mixed indices (hkl). (b) zero for mixed indices (hkl). (c) non-zero for un-mixed indices (hkl). Comparing X-ray diffraction from BCC and FCC structures, we see that (a) all reflection (hkl) does not show a measurable intensity. (b) Ihk0 = 0, for BCC. (c) FCC shows uniformly higher intensities than BCC. The structure factor for (a) HCP crystals is always an imaginary number. (b) CsCl is always a real number. (c) CsCl would be the same as a BCC crystal if Cs and Cl are randomly distributed in the unit cell. (d) NaCl can be derived from that of an FCC structure. A centrosymmetric crystal has (a) a point of inversion symmetry at the origin. (b) a structure factor that is always a real number. (c) a diffraction pattern that is also centrosymmetric. The Friedel law (a) states that non-centrosymmetric crystals also have a centrosymmetric diffraction pattern. (b) reduces the 32 point groups in three dimensions to the 11 Laue groups. (c) has no relevance to diffraction in theory or practice.

OUP CORRECTED – FINAL

476

X-Ray Diffraction 14. In diffraction, the presence of glide planes (a) gives rise to systematic absences. (b) and/or screw axes do not make any difference. (c) and/or screw axes have well-defined absences. 15. In diffraction, since we only measure the intensities (a) we do not know the phase of the diffracted waves. (b) we can always determine the position of all atoms in the unit cell. (c) we cannot unequivocally determine the position of all atoms in the unit cell. (d) we can determine the position of all atoms in the unit cell of centrosymmetric crystals. 16. The broadening of a diffraction peak is (a) affected by crystallite or grain size. (b) affected by microstrain in the individual crystallite. (c) large for single crystal specimens. (d) described by the Scherrer equation. (e) affected by instrumentation factors. (f) always proportional to sec θ. (g) equivalent to extending the infinitesimal reciprocal lattice point to a finite-sized node. (h) not interpretable in the Ewald sphere construction. 17. A collimator is used to produce ___________ of X-rays. (a) a monochromatic beam (b) a narrow parallel beam (c) only a circular shaped beam 18. A monochromator ________________ of X-rays. (a) produces a select and narrow wavelength (b) has as its main function the production of a narrow, well-shaped beam (c) works by absorbing a certain range 19. What we call a superlattice diffraction peak __________ in order–disorder transitions. (a) is a misnomer (b) is the result of the generation of a new lattice, different from the 14 Bravais lattices, (c) reflects the change from one Bravais lattice to another 20. A miniaturized X-ray diffractometer (a) can be constructed based on the Debye–Scherrer camera design. (b) is too fragile to survive the journey to Mars. (c) produced exceptional structural data of the minerals on the surface of Mars. 21. Chemical order–disorder transitions in alloys (a) cannot be probed by XRD. (b) can be studied by XRD for ALL materials.

OUP CORRECTED – FINAL

Exercises 477 (c) can be studied by XRD if the superlattice reflection intensity is significant. (d) cannot be studied by neutron diffraction. (e) shows a characteristic transition temperature that can be determined by XRD. 22. The Powder Diffraction Files (PDFs) contain entries (a) for more than 30,000 separate materials. (b) on the crystal structure and the three principal XRD lines, with relative intensities, for all materials. (c) that are too complicated to interpret or most routine work. 23. For polycrystalline materials with large unit cells (a) indexing of powder diffraction data is easy. (b) indexing of powder diffraction data is difficult because of overlapping peaks in the spectrum. (c) refinement of XRD data using a model structure often gives results comparable to single crystal X-ray analysis. 24. A crystal monochromator (a) works on the principle of Bragg diffraction. (b) can only be used for laboratory X-ray tubes, but not for synchrotron radiation. (c) can be designed to minimize the contribution from the Lorentzpolarization factor.

B. Problems 1. Following the example of NaCl, Example 7.5.3, derive the expression for the structure factor of the diamond structure that can be regarded as two interpenetrating FCC lattices shifted by the translation vector, τ = 1 4 (x + y + z). Further, show that Ihkl = 0 for mixed indices, and Ihkl = 0 when h + k + l = 4n + 2, where n is an integer. 2. Using Cromer–Mann parameters (Table 7.3.2) for Cr, calculate the structure factor for the following reflections: 000, 001, 011, 020, 121, 130, and 321. Check your results with the data in Example 7.5.2. 3. Using the Cromer–Mann parameters (Table 7.3.2) for NaCl, show that the structure factors for the following reflections are F000 = 112 F110 = 0 F002 = 85.7 F111 = 18.5 F220 = 72.8 ˚ and the lattice parameter for NaCl, a0 = 0.564nm. Use λ = 1.54 A

OUP CORRECTED – FINAL

478

X-Ray Diffraction 4. Consider a cubic unit cell of lattice parameter a ∼5 Å, subject to an X-ray diffraction experiment using Cu Kα radiation. i) What is the volume of the unit cell in real space? ii) What is the volume of the reciprocal unit cell? iii) What is the total available volume within the limiting sphere? iv) From i–iii, approximately how many independent reflections do you expect to see in a typical Debye–Scherrer pattern? v) Including Friedel law, how many independent rings of intensities do you expect to see? vi) Comment on the implications of the number of observable rings on a Debye–Scherrer pattern. Think about the potential overlaps. 5. Sphalerite, or zinc blende (§4.1.6), with the chemical formula of ZnS, has an FCC structure with a lattice parameter a = 0.541 nm. i) Write down the coordinates for Zn and S, in the unit cell. ii) Calculate the structure factor for sphalerite iii) For what values of hkl would you see a nonzero intensity? If possible, write this down in some simple form. iv) Calculate the intensities for the top three diffraction lines. Compare your results with PDF file. 6. A screw axis is normally represented by a 4 × 4 transformation matrix. Consider the 41 screw axis, parallel to the c-axis, which moves an atom in any general position (x, y, z) to three other equivalent positions: (–y, x, z+1/4), (–x, –y, z+1/2), and (y, –x, z+ ¾). i) Write down the structure factor for these four positions. ii) Now consider only the reflections of the type (00l) and rewrite the simplified structure factor. iii) Using the following mathematical relationships 

jl

eπi 2 =

j

and lim

1 − e2πil l

1 − e2πi 2

1 − e2πil

l→4n 1 − e2πi 2l

=4

show that reflections of the type (00l), for l = 4n, are absent. 7. For the Co–Pt multilayer data, using Cu Kα radiation shown in Figure 7.9.12, calculate the period from both the low-angle and high-angle scans. Comment on the similarity and/or differences in the result.

OUP CORRECTED – FINAL

Exercises 479 8. For the CuAu ordered structure shown in Figure 7.11.2, calculate the structure factor for the fundamental and superlattice reflections. Assume that the ordered alloy remains cubic with no change in lattice parameter. What is the intensity ratio Is /If ? Is our assumption valid in practice? 9. Order–disorder transition: Consider the (Cu0.5 Fe0.5 ) Pt3 alloy diffraction data shown in Figure 7.11.3. i) Calculate the structure factors and the ideal intensity ratios for the fundamental (111) and superlattice (100) reflections. ii) For the data at 303 K, approximate the integrated intensities, Is and If . iii) Calculate the order parameter using (7.11.9). iv) The chemical order parameter, S, can be related to Tc and the chemical ordering energy, Vorder , that stabilizes the ordered state, by the Bragg–Williams theory. Thus, near the ordering temperature 3 (Tc − T ) T = 4kB Tc

S2 = Vorder

If possible, calculate Tc , from the experimental dependence of the order parameter near the critical temperature. Then, use the value of Tc to calculate the chemical ordering energy. 10. Suppose you are given a sample and told that it is orthorhombic. You are also given the lattice parameters a, b, and c. Now you are asked to determine to which of the four orthorhombic lattices—primitive, basecentered, body-centered, or face-centered, as shown in Figure 7.Pr.10— this sample belongs.

c b a Primitive

Base-centered

Body-centered

Face-centered

Figure 7.Pr.10 The four orthorhombic lattices. Describe, in detail, how you would answer this question, including the indices of the peaks etc. that you will use in your analysis? Hint: Calculate the structure factors for each of the lattices.

OUP CORRECTED – FINAL

480

X-Ray Diffraction 11. SrZrO3 has the perovskite structure. i) What are the fractional coordinates (basis) of the three elements in a primitive cubic lattice. ii) Show that the X-ray structure factor for the (00l) reflection is given by   S00l = fSr +(–1)l fZr + 1 + 2(–1)l fO , where f are the atomic scattering factors. iii) Neglecting the angular dependence of the atomic scattering factor, calculate the ratio I(002) /I(001) of the X-ray diffraction intensities (Hint: Use Z ).

OUP CORRECTED – FINAL

Diffraction of Electrons and Neutrons

8 8.1 Introduction

482

8.2 The Atomic Scattering Factor for Electrons

482

79.2 29°

8.3 Basics of Electron Diffraction from Surfaces

485

63.43

8.4 Surface Electron Diffraction Methods and Applications

489

8.5 Transmission High-Energy Electron Diffraction

501

8.6 Transmission Electron Diffraction Methods

514

8.7 Examples of Transmission Electron Diffraction of Materials

527

8.8 Interactions of Neutrons with Matter

535

°

58. 7° 7.3

3

31.72° 63.43°

36°

79.2°

58.29°

37.38°

Transmission electron diffraction patterns taken from a single grain of the icosahedral phase in an Al-14at%Mn alloy and rotated through angles corresponding to the symmetry elements of the icosahedral group, m3 5, as shown in the stereographic projection (top left). For his discovery of these phases, also known as quasicrystals, Professor D. Shechtman was awarded the Nobel Prize in Chemistry in 2011. Moreover, this electron diffraction work fundamentally altered the definition of a crystalline material. The International Union of Crystallography changed the definition of a crystal from “a substance in which the constituent atoms, molecules, or ions are packed in a regularly ordered, repeating threedimensional pattern” to a broader one as “any solid having an essentially discrete diffraction pattern.” For further details, see the 2011 Nobel Prize press release at https://www.nobelprize.org/prizes/chemistry/2011/press-release/.

Principles of Materials Characterization and Metrology. Kannan M. Krishnan, Oxford University Press (2021). © Kannan M. Krishnan. DOI: 10.1093/oso/9780198830252.003.0008

Summary

540

Further Reading

542

References

543

Exercises

544

OUP CORRECTED – FINAL

482 Diffraction of Electrons and Neutrons

8.1 Introduction This chapter builds on the previous one and discusses diffraction of electrons and neutrons, with emphasis on the former. In general, because of their negative charge, the scattering of electrons by atoms is significantly stronger than that of X-rays. As a result, electron diffraction can be sensitive to surface crystallography, as well as probe the structure of very small volumes of materials in transmission using focused beams. The chapter begins by discussing the reciprocal lattice nets of surfaces and the Ewald sphere construction (§4.3.2) for diffraction in two dimensions. We then discuss two important surface diffraction methods: low-energy electron diffraction (LEED) and reflection high-energy electron diffraction (RHEED), illustrating each one with an important application in materials science. This is followed by a discussion of the basic principles of transmission electron diffraction (TED) in the context of their implementation in a transmission electron microscope (TEM), typically operating at high acceleration voltages (>100 kV). We restrict our discussion to the kinematical theory of electron diffraction, mostly in the column approximation, and for completeness, briefly introduce the dynamical diffraction of electrons. Diffraction from perfect crystals is treated in this chapter, and studies of imperfect crystals are introduced in §9.5.2. Practical implementation of TED includes three important methods: selected area diffraction (SAD), Kikuchi lines and patterns, involving both elastic and inelastic scattering, and convergent beam electron diffraction (CBED) using focused probes. These three are introduced and some typical applications of TED are presented. We conclude with a brief introduction to neutron scattering, including their use in studies of magnetic order. The emphasis in this chapter is on electron diffraction as they are readily encountered in most institutions and laboratories and unlike neutron sources that are generally available only at few national user facilities. For those interested in pursuing these subjects further, a more detailed discussion of TED can be found in numerous texts at different levels of sophistication. The list of good sources is too long, but a partial selection in chronological order of publication is Hirsch et al. (1965), Thomas and Goringe (1980), Cowley (1984), Williams and Carter (1996), and Fultz and Howe (2013). Similarly, surface electron diffraction is discussed in Pendry (1974) and Vickerman and Gilmore (2009), and neutron scattering is discussed in Bacon (1962), Schwartz and Cohen (1987), and Willis and Carlile (2009).

8.2 The Atomic Scattering Factor for Electrons Unlike X-rays, electrons interact with both the nucleus and the surrounding electron cloud of the atom; the nature of these two interactions is opposite in sign

OUP CORRECTED – FINAL

The Atomic Scattering Factor for Electrons 483 because of their respective charges. The scattering of an electron, considered as a particle with charge, e– , and wavelength, λ (=h/mv), by the nucleus, considered as a point charge, q (= +Ze), can be analyzed based on their electrostatic interaction. This analysis is similar to the classic Rutherford scattering of positively charged α-particles (He2+ , +2e) by the atomic nuclei, which was introduced in §5.3.5.3. In the scattering of electrons by an atom, the α-particle in Rutherford scattering is replaced by the electron (Fig. 8.2.1) and the fraction of electrons, dσ , scattered into a solid angle, d, at an angle, 2θ , by the nucleus, is given by    dσ 1/2 Ze2  = | fe (2θ)| =   d 2mv2 sin2 θ

(8.2.1)

where fe (2θ ) is the (complex) electron scattering amplitude or scattering factor. The derivation of (8.2.1) can be found in texts on modern physics, such as Sproull and Phillips (2015). In the first Born1 approximation, i.e. assuming each incident electron is scattered only once within the atom, only the amplitude of fe (2θ ) is important for scattering and its phase is neglected. However, the phase component of fe is important in high resolution phase contrast imaging in a TEM, as discussed in §9.2.7.3. Note that the Coulomb scattering depends on the product of two charges (and the scattering angle). In addition to the positively charged nucleus, the electrons are also scattered by the negatively charged atomic electron cloud. It is logical to assume, as we have done in §7.3 for X-rays, that the scattering from all the surrounding electrons is given by the sum of their individual scattering, provided we account for the phase factor arising from the spatial distribution of the electronic charge cloud around the nucleus. The latter has been calculated for incident X-rays as the atomic scattering factor, fatom , (7.3.8). Now, if we assume that the electron distribution around the nucleus is spherically symmetric, we can replace the electron cloud by a point charge, –fatom e, at the center of the nucleus, and modify (8.2.1), to calculate the total electron scattering factor, fe , for the atom as a sum of the nuclear and electron scattering contributions: fe =

(Z − fatom ) e2 2mv2 sin2 θ

=

me2 λ2 2h2 sin2 θ

(Z − fatom )

dσ u

2θ +Ze

d Figure 8.2.1 Rutherford scattering geometry of electrons with atomic nuclei, +Ze, and impact parameter, u. Electrons originating from an area dσ are scattered through an angle 2θ into a solid angle, dΩ. The scattering angle depends on the impact parameter; as u decreases, 2θ increases. For small angle scattering, particularly relevant for transmission electron diffraction, the differential scattering cross section, dσ /dΩ ∝ θ –4 .

(8.2.2)

where the de Broglie wavelength, λ = h/p, (1.3.3), is used for the electron. Substituting the numerical values for the physical constants, we get  fe

sin θ λ

 = fe (s) =

0.0239 (Z − fatom ). s2

(8.2.3)

in units of Å, unlike fatom for X-rays, which is dimensionless. Note that the scattering parameter, s, is defined as in §7, i.e. s = sinλ θ . Strictly speaking, the relativistic mass of the electron increases with velocity as m = γ m0 , where m0 is

1 Max Born (1882–1970) was a German physicist who received the 1954 Nobel prize in Physics and was cited for “fundamental research in Quantum Mechanics, especially in the statistical interpretation of the wave function.”

OUP CORRECTED – FINAL

484 Diffraction of Electrons and Neutrons its rest mass and γ is the relativistic mass correction factor. Thus, for electrons with energy, E (keV), it is necessary to multiply (8.2.3) by γ, given by

(a)

f

γ= 

(b)

(c)

sinθ λ Figure 8.2.2 Schematic representation of the variation of the atomic scattering factors for (a) electrons, (b) X-rays, and (c) neutrons, as a function of the scattering parameter, sinλ θ . Notice that neutron scattering has a negligible angular dependence.

1 1 − (v/c)

2

=1+

E E [keV ] 1+ me c2 511

(8.2.4)

to obtain practically useful values of electron scattering factors. Note that the constant, 0.0239, in (8.2.3) is much larger than the scattering of X-rays by a single electron. Further, (Z − fatom ) /s2 > fatom for s ≤ 0.3 − 0.4, and (Z − fatom ) /s2 < fatom for s > 0.3 − 0.4. Figure 8.2.2 shows the relative atomic scattering factors for X-rays, electrons, and neutrons, as a function of the scattering parameter, s. Most importantly, because fe > fatom , for values of s encountered in TED, the scattering of electrons by atoms is greater than that for X-rays by a large factor (103 –104 ). This has a number of important implications for the diffraction of electrons by materials. First, scattering effects from small volumes or very small quantities of materials can be readily detected, making it possible to undertake SAD (such as in a TEM, §8.6 and §9.2.7.1), and diffraction studies with electron probes focused on the nanoscale. Second, electron beams are attenuated much more rapidly and are much less penetrating than X-ray beams in materials. As a result, electron beams are readily absorbed/scattered, even by air, and require vacuum columns for diffraction experiments. Third, electron diffraction in transmission requires very thin specimens (typically, ≤100 nm thick) to avoid multiple scattering, and in the parlance of TEM, suitable specimens for observations are often referred to as thin foils. Fourth, the decrease of diffracted intensities with the scattering angle, 2θ, is much more rapid for electrons compared to X-rays (Fig. 8.2.2). Combined with much shorter wavelengths (than X-rays), and a much larger radius for the Ewald sphere (Fig. 4.3.2) it results in most of the TED patterns being largely scattered in the forward direction and restricted to a much smaller angular range (2θ ≤ ±4◦ ). Fifth, for thicker specimens, electron diffraction is often carried out in a glancing geometry, but such electron diffraction patterns (RHEED) provide information only from the surface structure (t ≤ 10 nm) of the specimen. RHEED is discussed in more depth in §8.4.3. Alternatively, surface (monolayer) structures can be studied by using low-energy (∼100 eV) electrons, and LEED is discussed further in §8.4.1. In general, electrons after scattering once can be easily diffracted multiple times by different lattice planes over short distances in the material. In contrast, the probability of X-rays being scattered multiple times is small and can be ignored. Hence, the simple kinematical approach where the intensity of the diffracted beam is considered to be proportional to the square of the structure factor, i.e. Ihkl = |Fhkl |2 , valid for X-rays, (7.4.8), may generally give erroneous results for electron diffraction intensities. Multiple scattering effects, or dynamical diffraction

OUP CORRECTED – FINAL

Basics of Electron Diffraction from Surfaces 485 theory, is required to describe electron diffraction accurately; however, for very thin films, the kinematical theory will suffice as a good approximation. The dynamical theory of electron diffraction is introduced very briefly in §8.5.4, but a detailed discussion is well beyond the scope of this book. Interested readers are referred to more specialized texts, e.g. Hirsch et al. (1965), Cowley (1984), and Fultz and Howe (2013). The discussion in §8.3 follows from the introduction to crystallography and diffraction presented in §4. In particular, we revisit the basics of the “crystallography” of surfaces and the related modification in two dimensions of the Ewald sphere construction. Example 8.2.1: Calculate the atomic scattering factor for FCC Cu [111] reflection for electrons at 5 keV and 100 keV incidence. The lattice parameter aCu = 3.62 Å. Solution: The interplanar spacing d 111 = 2.08 Å, for Cu from Table 4.1.2. The deviation parameter s111 = sinλ θ = 2d1111 = 0.239. Using the value of s111 , we determine fatom = 22.094, following (7.3.10) and the Cromer–Mann parameters for Cu in Table 7.3.2. Applying (8.2.3), we get fe (s111 ) = 2.884 Å. Applying the relativistic mass correction factor, (8.2.4), we get fe (5 keV) = 2.912 Å fe (100 keV) = 3.448 Å

8.3 Basics of Electron Diffraction from Surfaces In Chapter 7, we described the use of X-ray diffraction in the bulk for threedimensional crystal structure measurements. Since X-rays are only scattered weakly by the electrons surrounding the nucleus, they penetrate deeply into materials and allow us to probe the bulk crystal structure. Neutrons, as we will see in §8.8, interact even more weakly with solids, penetrate more deeply and with negligible angular dependence (Fig. 8.2.2). Hence, X-ray and neutron scattering are not the first choice for surface structure determination, although such measurements are possible using specialized geometries (see §7.8.4 on Xray reflectivity). Thus, electrons that are strongly interacting, with mean free path lengths of the order of a few nanometers at energies of ∼100 eV, are the logical choice for diffraction studies of surfaces. Recall that in two dimensions there are only five lattice nets possible (Fig. 4.1.15) as opposed to the 14 Bravais lattices (Table 4.1.1) in three

OUP CORRECTED – FINAL

486 Diffraction of Electrons and Neutrons dimensions. We now begin a discussion of surface diffraction with a brief summary of the nomenclature used to describe surface lattices, followed by the Ewald sphere construction, now modified for two dimensions.

8.3.1 Surface Reconstruction, Surface Nets, and Their Notation Ideally, we can identify a surface by referring to the terminating plane of the bulk material, i.e. Au(100), Fe(110), etc. Mathematically, we can describe the surface net as a translation vector T = mas + nbs (b)

(a)

(8.3.1)

(c)

(d)

bs as

(1 × 1)

01

00

(2 × 1)

11

10

01 11 11 2 00 1 0 10 2

p(2 × 2)

01

b*s

c(2 × 2) (√2 ×√2 ) R45°

11

01

11 22 00

11 11 22

10

00

10

a*s

Figure 8.3.1 Examples of surface lattice structures (top row), where the filled dots represent the periodicity of the substrate and the crosses represent the surface lattice: ( a) (1 × 1) reconstruction, (b) (2 × 1) reconstruction, (c) a primitive (2 × 2) reconstruction, and (d) a (2 × 2) reconstruction with an additional centering translation that give rise to a multiple cell—the alternative unit cell, also shown, is rotated by 45◦ . Both (c) and (d) show adsorbate atoms on the surface; if the adsorbate were to and the substrate an Au(100) surface, the nomenclature for (d) would be √be oxygen, √  2 × 2 R45◦ − O. In all cases, the substrate lattice nets are shown by dashed lines and explicitly modified as Au(100) the surface lattice net is shown by continuous lines. For each reconstruction in the top row the corresponding reciprocal lattice, including the unit cell, is shown in the bottom row; typically, these patterns are observed in LEED, for normal electron incidence. See Figure 8.4.3. Adapted from Woodruff and Delchar (1992).

OUP CORRECTED – FINAL

Basics of Electron Diffraction from Surfaces 487 where the subscript, s, refers to the surface, m and n are integers, and as and bs are surface unit cell vectors of the crystal (or positions occupied by surface adsorbates). In reality, due to the break in symmetry at the surface, the atoms on the top-most monolayer of the surface can rearrange themselves into a new net. Physically, this surface reconstruction arises from a minimum energy configuration different from the bulk because the break in symmetry reduces the number of nearest neighbor atoms on the surface. This description also applies to other atoms adsorbed on surfaces. In the general case, the reconstructed surface unit-cell vectors, as and bs , are related to the translation vectors, a and b, of the ideal bulk terminating layer as as = Ma and bs = Nb

(8.3.2)

where M and N are integers. The reconstructed surface is then known as a (M × N) reconstruction. Further, if the surface net is rotated by φ, with respect to the underlying bulk lattice, the nomenclature is modified as (M × N)Rφ (Fig. 8.3.1d), where R indicates rotation. Moreover, if the surface net is best defined using a centering translation (Fig. 8.3.1d) instead of it being a primitive one (Fig. 8.3.1c) it is indicated explicitly as c(M × N). Finally, if the surface layer includes a specific adsorbed species, the adsorbate is explicitly indicated. Figure 8.3.1 provides some examples of surface nets, including reconstruction and positions occupied by the adsorbates, as well as the nomenclature used to describe them.

8.3.2 Reciprocal Lattice Nets and Ewald Sphere Construction in Two Dimensions The reciprocal lattice construction for a two-dimensional surface lattice follows from our definition in three dimensions, (4.2.1), but in a simplified form as a∗s =

2π  bs ⊗ nˆ A

b∗s =

2π  nˆ ⊗ as A

(8.3.3)

 where the area, A = as · bs ⊗ nˆ , and nˆ is a unit vector normal to the surface. Figure 8.3.1 also shows the corresponding reciprocal lattice nets (bottom row). Diffraction processes from surfaces (Fig. 8.3.2a) can also be described by the Ewald sphere construction (Fig. 4.3.2) but appropriately modified for the twodimensional periodicity parallel to the surface (Fig. 8.3.2b). As a consequence, only the wave vector component parallel to the surface is conserved with the addition of a reciprocal lattice vector. We follow our earlier convention for incident, k0 , and diffracted, k, beams, but then use superscript, ||, and ⊥ to indicate their components parallel and perpendicular to the surface.

OUP CORRECTED – FINAL

488 Diffraction of Electrons and Neutrons (a) k0 k

φ

Ψ

d

k

Figure 8.3.2 Schematic representation of surface diffraction in (a) real space, where φ is the angle of incidence, and ψ is the exit angle. Each cross (dot) represents a row of surface (bulk) atoms normal to the page. The corresponding Ewald sphere construction in reciprocal space, including the reciprocal lattice rods, is shown in (b); notice the two possibilities for the diffracted wave, with one, shown dashed, propagating into the crystal that is not observed. Adapted from Vickerman and Gilmore (2009).

2π d

(b)

k

Reciprocal Lattice Rod

g

k0 Surface k

– (04)

– (03)

– (02)

– (01)

(00)

(01)

First, the conservation of energy gives k20 = k2

(8.3.4)

or, using the parallel and perpendicular components, we get  2  2  2  2  k0 + k⊥ = k + k⊥ 0

(8.3.5)

Similarly, the conservation of momentum gives 

k = ghk + k0

(8.3.6)

OUP CORRECTED – FINAL

Surface Electron Diffraction Methods and Applications 489 where ghk = ha∗s + kb∗s is a surface reciprocal lattice vector defined in (8.3.3). Strictly speaking, the perpendicular component, k⊥ , need not be conserved in the surface diffraction process. In addition, now that the diffraction conditions are dependent on a reciprocal lattice-net vector with only two components, we can also represent the diffracted beam with only two indices, (hk). Further, the Ewald sphere construction is modified (Fig. 8.3.2b). The incident and diffracted beams lie in three dimensions (Fig. 8.3.2a) but, most importantly, reciprocal lattice rods of infinite length perpendicular to the surface and passing through the reciprocal lattice points replace the reciprocal lattice of Figure 4.3.2. This is because the surface is a two-dimensional net with infinite periodic repeat distance normal to the surface. As the distance between adjacent reciprocal lattice points is inversely proportional to the corresponding distance in real space, we expect the reciprocal lattice points normal to the surface to be infinitesimally close to one another, and to be effectively perceived as a continuous line or rod. The Bragg diffraction condition is satisfied for every incident beam that emerges in a direction given by the intersection of the Ewald sphere with one of the reciprocal lattice rods (Fig. 8.3.2b). Compared to three dimensions, the presence of the reciprocal lattice rods for surfaces greatly relaxes the conditions for the observation of diffracted beams. In the bulk, a small change in incident electron energy (∝ radius of the Ewald sphere), or the direction of the scattering wave-vector, k, will results in the absence of many beams and the excitation of new ones. In contrast, for the case of surfaces, diffracted beams may occur at all energies provided the corresponding reciprocal lattice rod lies within the Ewald sphere. In addition, for each reciprocal lattice rod two diffraction beams can satisfy the Bragg diffraction conditions (Fig. 8.3.2b); however, only half of them are back-scattered and observed; the other half propagates forward into the bulk crystal and are not observed in surface diffraction. Finally, the indexing of the diffracted beams from surfaces is, by convention, with reference to the unreconstructed substrate real and reciprocal lattice net. As a result, if the reconstructed surface, or the distributions of adsorbates, have larger periodicities, the corresponding surface reciprocal lattice net would be smaller than that for the bulk lattice and the additional reciprocal lattice points are indexed with fractional indices (Fig. 8.3.1c,d). With this brief background of the diffraction of electrons from surfaces, we introduce two practical and commonly used surface diffraction methods involving low- (LEED) and high- (RHEED) energy incident electrons.

8.4 Surface Electron Diffraction Methods and Applications We can see from (1.3.5) that electrons of energy ∼150 eV have a wavelength of ∼0.1 nm (1 Å), making them suitable for diffraction from crystalline materials with similar lattice parameters. Further, from the universal curve (Fig. 1.3.6) plotted for the mean free path length of electrons as a function of energy, we can

OUP CORRECTED – FINAL

490 Diffraction of Electrons and Neutrons see that the penetration depth is ∼0.25 nm, of the order of a surface monolayer or two, for electron energies in the range 50–100 eV. This ensures that these lowenergy electrons have good surface sensitivity in diffraction experiments. When electrons of such low energy, incident normal to a crystal surface, are elastically back-scattered, the resulting diffraction pattern forms the basis of the technique of LEED and is used for surface structure determination. Alternatively, we can use high-energy electrons incident at grazing angles on the surface of a crystal. The shallow angle of incidence of the electrons results in a negligible component of their angular momentum normal to the surface. As a result, their penetration depth into the surface is very small, and their scattering from the surface in the reflection geometry forms the basis of RHEED. These two techniques, LEED and RHEED, along with a representative application for each one, are introduced in the sections to follow.

8.4.1 Low-Energy Electron Diffraction (LEED) Figure 8.4.1 shows a schematic diagram of an experimental set up for LEED measurements. The electron gun, described in §5.2.2, produces a beam of variable energy in the range ∼20–300 eV, an energy spread of ∼0.5–1.5 eV, and a current of ∼1 μA. The beam is delivered to the specimen surface by a system of electrostatic focusing lenses. After interaction with the specimen surface, the back-scattered electrons travel in straight lines in the field-free region to a set of three grids surrounding the electron gun. The first grid is set at ground, the same potential as the specimen. The second set of grids are set to discriminate and eliminate any inelastically scattered electrons that have lost small amounts of energy, and allow only the nearly elastically scattered ones to go through. Such a discriminator to remove electrons that have lost more than a few eVs of energy is also known as a retarding field analyzer. The final grid is set at +5 kV to reaccelerate the diffracted electrons and have them impinge on a fluorescent screen and create the LEED pattern. Electron Gun, Potential ~ –Ve

+ 5kV Figure 8.4.1 A schematic representation of the LEED apparatus. A lowenergy beam is incident normal on the specimen and a hemispheric analyzer, with multiple grids for energy discrimination, detects the electrons diffracted from the surface in the backward direction.

Fluorescent Screen –Ve + V

Grids Specimen

OUP CORRECTED – FINAL

Surface Electron Diffraction Methods and Applications 491

–1 0

–2

1 LEED Screen

2

–3 3

Ewald Sphere

Specimen

30

20

11

01

11

12

10

00

10

20

01

11

30

LEED Pattern

Figure 8.4.2 shows the corresponding Ewald sphere construction for normal incidence. Comparing the two figures, it is clear that the LEED pattern observed is just a projection of the surface reciprocal lattice net, with a magnification determined by the energy of the incident electrons. Note that, strictly speaking, the position of the gun eliminates the possibility of recording the (00) beam for normal incidence. Further, LEED experiments are extremely sensitive to the cleanliness of the specimen surface; typically, experiments are carried out in an ultra-high vacuum (UHV) system (∼10–10 torr, or ∼10–8 Pa) giving rise to atomically clean surfaces. In a typical LEED experiment a portion of the reciprocal space, i.e. those electrons that are diffracted only in the backward direction are recorded. The most common data acquired is the variation of the intensity, I (hk), as a function of position along a small portion of the Ewald sphere (Fig. 8.4.2) where h and k, are the diffraction indices. The positions of the spots and their relative intensities (e.g. Fig. 8.4.3) give information on the surface or over-layer unit lattice net size and shape. The exact positions of the atoms on the surface cannot be accurately determined from a visual inspection of the diffraction pattern, but instead, it requires an analysis of the diffraction intensities. The latter is more complicated to do as it requires that the intensity distribution, I (V ), or the variation of the

Figure 8.4.2 The Ewald sphere construction for LEED under normal incidence. A cubic (001) surface, a schematic of the two-dimensional LEED pattern, typically observed (see Fig. 8.4.3a). The diffracted beams, corresponding to a linear section of the reciprocal lattice, detected in the backward direction are indicated in the Ewald sphere construction.

OUP CORRECTED – FINAL

492 Diffraction of Electrons and Neutrons intensity as a function of incident beam energy/voltage for a number of diffraction geometries, be measured and compared with model calculations (see Pendry, 1974). Other structural information, including surface phase transitions can be obtained by measuring the intensity variation, I (T ), as a function of temperature. Further, the kinetics of order–disorder transitions (§7.11.4) on surfaces can also be obtained by measuring the time evolution of diffraction intensities (Lagally, 1985). Alternatively, the intensity variation, I (α), of specific diffraction features as a function of adsorbate surface coverage, α, is very useful in adsorption studies. A simple example of the adsorption of different elements on a W(100) surface is presented in the next section. Last, but not least, LEED is commonly used to establish the cleanliness and surface structural order; the latter is also correlated with other physical properties such as lattice dynamics, surface diffusion kinetics, and electronic and magnetic properties. A quick summary of LEED can also be obtained in the Encyclopedia of Materials Characterization (Brundle, Evans, and Wilson, 1992), or in Amelinckx, Gevers, and van Landuyt (1978); for more detailed discussions, see Pendry (1974) and/or Lagally (1985). Example 8.4.1: Plot the expected LEED pattern for the two cubic (110) surfaces shown in the following figure for (a) smooth and (b) stepped surfaces. For (b) assume that the coherence length is larger than the step width. Comment on what will happen if the step width is irregular. (b)

(a)

Surface

Smooth Stepped

LEED Pattern

Solution: (a) The reciprocal lattice for the (110) surface can be calculated from (8.3.3). Note that the reciprocal lattice net is rotated by 90◦ with respect to the real lattice.

OUP CORRECTED – FINAL

Surface Electron Diffraction Methods and Applications 493

(b)

Because of the periodic steps, the surface repeat unit is now enlarged by a factor of four leading to a LEED pattern that is roughly a (4 × 1) reconstruction. Strictly speaking the real repeat units should be (42 + 12 )1/2 = (17)1/2 . If the step width is irregular, the additional spots will be streaked in the direction of the disorder.

8.4.2 Adsorption Studies on Surfaces Using LEED Figure 8.4.3 shows a set of LEED patterns for various adatoms adsorbed on a W(100) surface. The diffraction pattern from a clean W(100) surface (Fig. 8.4.3a) corresponds to a surface with a square lattice in both real and reciprocal space in Figure 8.3.1a. When oxygen adatoms are adsorbed on the W(100) surface (Fig. 8.4.3b) additional spots in the half-order position, indicating a doubling of the surface lattice parameters for oxygen, consistent with the primitive p(2 × 2)–O superstructure is seen; this corresponds to Figure 8.3.1c. In contrast, the LEED pattern (Fig. 8.4.3c) for the adsorption of hydrogen on the W(100) surface is nonprimitive. It is clearly a centered, c(2 × 2)–H superstructure (Fig. 8.3.1d) with a rotation of 45◦ with respect to the substrate, W(100), surface. Finally, Figure 8.4.3d shows the LEED pattern for the co-adsorption of carbon monoxide and nitrogen on a W(100) surface. Clearly, the pattern corresponds to a (4 × 1)–N, CO reconstruction. Further, when compared to the (2 × 1) pattern in Figure 8.3.1b, this pattern confirms the presence of two domains of the (4 × 1) superstructure rotated by 90◦ with respect to each other. From this brief illustrative example, we conclude that LEED provides a very good map of the reciprocal lattice net of the specimen surface monolayer, and allows the determination of the surface lattice vectors, as and bs , and

(a)

(b)

(c)

(d)

Figure 8.4.3 Photographs of LEED diffraction patters from (a) (1 × 1) or W(100) surface, (b) p(2 × 2)–O reconstruction for a monolayer of oxygen atoms adsorbed on the surface, (c) a non-primitive c(2 × 2)–H surface, and (d) two domains of (4 × 1)–N,CO reconstructed surfaces of adsorbed gases on W(100) surface. Adapted from Amelinckx, Gevers, and van Landuyt (1978).

OUP CORRECTED – FINAL

494 Diffraction of Electrons and Neutrons their relationship to the substrate lattice vectors, a and b. Further, LEED patterns provide some information about the lateral, inter-atomic interactions between the adsorbed atoms on the surface. For example, if there were no interactions, the adsorbed atoms would be distributed randomly on top of the atoms on the terminating layer of the substrate; however, if there were to be a lateral repulsion between neighboring adatoms, the tendency would be to form a c(2 × 2) superstructure, where each adatom is separated by at least one substrate surface atom (Fig. 8.3.1d; Fig. 8.4.3c).

8.4.3 Reflection High-Energy Electron Diffraction (RHEED) Figure 8.4.4a shows a typical experimental arrangement for RHEED. The highenergy (5–100 keV), parallel beam is incident at grazing angle φ ∼ 87–90◦ (Fig. 8.4.4b) and the elastically diffracted electrons are detected by a retarding field analyzer and then displayed on a fluorescent screen. If necessary, the intensity of a specific diffracted beam, useful for measuring RHEED oscillations (Fig. 8.4.8) can also be monitored as function of time or film growth. The higher energy (smaller wavelength, λ, or larger wavevector, k0 ) of the incident beam in RHEED, compared to the much smaller energy in LEED, makes a significant difference in the observed diffraction patterns. The Ewald sphere in RHEED (Fig. 8.4.5a) is now very large compared to the reciprocal lattice dimensions. As a result, its intersection with the reciprocal lattice rods, for an Fluorescent Screen

(b)

(a) Fluorescent Screen

01

Specimen

00 θ

s

φ Electron Gun (5–100kV)

Specular Beam Spot

L ψ

01

Shadow Edge

Figure 8.4.4 (a) A schematic of the experimental apparatus used for reflection high-energy electron diffraction in side view. (b) A three-dimensional view of the same, defining the incident angle, φ ∼ 87–90◦ , the azimuth angle, ψ, and the diffraction angle, θ , as well as other details of the experiment used in determining details of the surface lattice net from the RHEED pattern. Adapted from [1].

OUP CORRECTED – FINAL

Surface Electron Diffraction Methods and Applications 495 (a)

(b)

Ewald Sphere

Reciprocal Lattice Rods

Diffracted Beams, k

RHEED Spots

Reciprocal Lattice Rods

k Specimen

Incident Beam, k0 (c)

Ewald Sphere

k0 Specimen

(d)

Ewald Sphere

k0

φ 000

Reciprocal Lattice Rods

φ 000

Specimen

Reciprocal Lattice Rods

Figure 8.4.5 (a) The Ewald sphere and the reciprocal lattice rods involved in RHEED (shown in the inset). The scattering geometry, including the incident (k0 ) and scattered (k) beams, and the associated scattering vector ( k), is illustrated. (b) The ideal two-dimensional pattern from a Ge(001) surface. (c) For high-energy electrons incident at grazing angles, φ ∼ 90◦ , the Ewald sphere (with radius exaggerated here for clarity as a straight line) intercepts very few reciprocal lattice rods. (d) Reducing the angle of incidence allows few more reciprocal lattice-rods to be intercepted by the Ewald sphere. (b) Obtained courtesy of Prof. Tsui, University of North Carolina, Chapel Hill. Adapted from Vickerman and Gilmore (2009).

ideal two-dimensional specimen surface, appears as a set of closely spaced dots (Fig. 8.4.5b). The corresponding RHEED pattern shows streaks (Fig. 8.4.6a). In practice, the radius of the Ewald sphere is so large that only a few spots/streaks are observed in a RHEED pattern. However, if the angle of incidence, φ, is changed additional diffraction conditions may be satisfied (Fig. 8.4.5c,d) as other lattice rods may be intercepted by the Ewald sphere. Thus, the arrangement of reciprocal lattice rods in three dimensions can be obtained. Such changes in φ are obtained

OUP CORRECTED – FINAL

496 Diffraction of Electrons and Neutrons by rocking the specimen, or alternatively by rotating it (changing the azimuthal angle, ψ, in Figure 8.4.4b) about its normal. However, in practice, the surface tends to have atomic scale roughness either due to incomplete coverage or surface defects, and now the surface is called quasi two dimensional. The size of the surface unit mesh can be readily obtained from the RHEED pattern. From Figure 8.4.4b, with the separation, s, measured between the streaks of the pattern, and distance, L, between the specimen and the fluorescent screen, we obtain the diffraction angle, θ , as tan θ =

s ≈ sin θ L

(8.4.1)

for small θ . Further, for a cubic lattice (a = b), the surface diffraction satisfies the condition,  1/2 λ a sin θ = h2 + k2

(8.4.2)

where h and k are the diffraction indices. Then, substituting for sinθ from (8.4.1), we get 1/2  L   λ a = h 2 + k2 s

(8.4.3)

and thus, allowing the lattice parameter, a, to be determined. In RHEED, as well as in LEED, we are interested in knowing the coherence length/area involved in a measurement. By coherence length/area, we mean how large an area of the surface is involved in the production of a particular diffraction pattern/measurement. Both the energy spread of the incident beam and its angular divergence contribute to the coherence length of a RHEED pattern, and for most common experimental arrangements it is ∼200 nm; typically, it is much smaller for LEED (5–10 nm). In other words, RHEED may not be able to detect any surface disorder if it occurs on a scale larger than the coherence length. However, for the small grazing angles used, RHEED is very sensitive to surface disorder on scales much smaller than the coherence length (Fig. 8.4.6 a,b). In addition to the quasi-two-dimensional surface arising from surface disorder causing streaks, the primary RHEED beam may pass through any surface protrusions or asperities (Fig. 8.4.6b). Such transmission of the incident beam will also be sensitive to the additional periodicity of the lattice planes normal to the surface. As a result, instead of streaks, spots appear in the RHEED pattern with their separation being inversely related to the inter-planar spacing normal to the surface. The contribution from the reflected and transmission features can be easily resolved in RHEED. When the incident angle, φ, is varied the transmission features will change in intensity but not in position; in contrast, the surface features will move continuously in position because the intersection of the Ewald sphere with a lattice rod will move up or down as φ is changed.

OUP CORRECTED – FINAL

Surface Electron Diffraction Methods and Applications 497 (a)

Ewald Sphere Reflection

Single Crystal with atomic surface roughness –Quasi 2D

Diffracted Beam, k Incident Beam, k0

Shadow Edge

Straight through beam (b)

Transmission k k0

Single Crystal Islands –3D

Straight through beam (c)

Transmission k

Polycrystalline –3D

k0

Straight through beam

Figure 8.4.6 A schematic representation of three possible RHEED scattering geometries, depending on crystal surface morphology, with actual patterns obtained from metal layers grown on Ge(001). In each row, in sequence, from left to right we show a schematic of the film surface and indicate the incident and diffracted electrons in real space (left), the corresponding Ewald sphere construction, the resultant diffraction pattern (schematic) with both the shadow edge (dashed line) and the straight-through beam, which is the incident beam without hitting the substrate, and the observed RHEED pattern (right). The different morphologies are (a) a clean single crystal surface with atomic scale roughness, (b) single crystal surface with island growth, and (c) a polycrystalline film, where the grey sphere is the reciprocal structure for a polycrystalline crystal with randomly oriented grains. Courtesy Prof. Tsui, UNC, Chapel Hill. Adapted from [2].

OUP CORRECTED – FINAL

498 Diffraction of Electrons and Neutrons

8.4.4 RHEED Oscillations: In Situ Monitoring of Thin Film Growth Equilibrium growth morphologies of thin films can be predicted based on a simple free energy consideration [3] comparing the free energy of the substrate, γsubstrate , which exists before the film growth, with the sum of the free energies of the film surface, γfilm , and the film/substrate interface, γinterface , which is established after the film has been deposited. If γsubstrate > γfilm + γinterface

(8.4.4)

the first monolayer of the film will prefer to coat the entire substrate surface and a layer-by-layer growth is expected (Fig. 8.4.7i). However, once the first monolayer is deposited, further growth of the film will not encounter the substrate surface and the interface energy is now also different. Moreover, the lattice mismatch (if any) between the substrate and the film will give rise to a misfit strain energy, which will increase with growing film thickness. As a result, after the first layer, a modified form of growth may replace the continued layer-by-layer growth, where the next layer forms as islands. This is known as the Stranski–Krastanov (S–K) growth mode (Fig. 8.4.7ii). This mode not only allows the energetically favorable first layer of the film to remain exposed but also makes it possible for the subsequent islands to relieve any strain by lateral relaxation. Alternatively, if γsubstrate < γfilm + γinterface

(8.4.5)

the film nucleates directly as a three-dimensional island (Fig. 8.4.7iii), and leaves as much of the low-energy substrate surface exposed as possible. (a) Equilibrium Growth Modes

Figure 8.4.7 (a) Equilibrium growth mode determined by different surface and interface energies. (i) Layer-bylayer growth. (ii) Stranski–Krastanov (S–K) growth and (iii) island growth. (b) Non-equilibrium growth modes are a function of substrate temperature, evaporation rates, and step density (iv) step-flow growth and (v) island growth. Adapted from Krishnan (2016).

Substrate

Substrate

Substrate

(i) Layer by layer

(ii) Stranski–Krastanov

(iii) Island

(b) Nonequilibrium Growth Modes

Substrate

Substrate

(iv) High step density High substrate temperature Low deposition rate

(v) Low step density Low substrate temperature High deposition rate

OUP CORRECTED – FINAL

Surface Electron Diffraction Methods and Applications 499

(a) Monolayer Growth [001]

Electron Beam

[110]

Surface Coverage

RHEED Intensity

α=0

[110]

α = 0.25 α = 0.5 α = 0.75

(b)

RHEED intensity (a.u)

α = 1.0

τ

0

1

2

GaAs (001)

4 5 3 Deposition time (s)

6

7

Equilibrium growth of thin films generally requires high substrate temperatures (T g > 200 ◦ C); however, nonequilibrium growth at lower temperatures (T g < 200 ◦ C) and high deposition rates are often employed. In such conditions, the role of surface steps becomes critical in determining nonequilibrium growth morphology [4]. At high growth temperatures and/or low deposition rates, the deposited atoms have sufficient time and energy to diffuse to the nearest step followed by bonding at this location with high coordination number (Fig. 8.4.7iv). This is known as step-flow growth. However, at reduced growth temperatures, or high growth rates, or low step densities, the deposited atoms do not have sufficient time and energy to diffuse to the step edges. Instead, when they arrive on the substrate surface they nucleate spontaneously as islands, which continue to grow by incorporating new adatoms as they are deposited (Fig. 8.4.7v). Eventually, they coalesce into a smooth monolayer, but then nucleate again as islands forming a rough, incomplete, monolayer surface. This process is repeated as the film grows and results in alternating rough and smooth surfaces for each additional half a

Figure 8.4.8 Variation of the intensity of the specularly reflected beam in RHEED as a function of the fractional surface coverage, α, for layerby-layer growth. (a) The intensity oscillates from maximum (α = 0), to a minimum (α = 0.5), to a maximum (α = 1.0) again. (b) RHEED oscillations with deposition time for films of GaAs(001) grown by molecular beam epitaxy (MBE), confirming the layerby-layer growth morphology. Adapted from [5].

OUP CORRECTED – FINAL

500 Diffraction of Electrons and Neutrons monolayer coverage. The most common way to monitor this type of growth is by RHEED, which shows a periodic oscillation of the intensity with every halfmonolayer coverage (Fig. 8.4.8). The scattering geometry is set up in such a way that the reflected electron beams from the upper terraces interfere destructively with the ones from the lower terraces. Thus ideally, this interference is out of phase, or completely destructive with zero intensity, for any half-monolayer coverage. Alternatively, it can be in phase, or constructive with maximum intensity, for any complete monolayer coverage. In practice, the beam intensity oscillates showing a minimum (maximum) for out-of-phase (in-phase) conditions, and these RHEED oscillations can not only be a very good measure of the growth morphology (Fig. 8.4.8), but compare very well with surface imaging methods (Fig. 8.4.9).

(b)

(c)

10 nm

40 nm

40 nm

200

200

STM

(a)

200

T = 20°C

T = 180°C

50

Intensity

100

0

150

150 Intensity

Intensity

150

100

1 2 3 Time (min)

4

0

100 50

50 0

T = 250°C

0

1 2 3 Time (min)

4

0

0

2 1 3 Time (min)

4

Figure 8.4.9 Images taken with a scanning tunneling microscope or STM (described in §11) of a Fe monolayer growing on an Fe(100) surface. Three different growth modes: (a) island, (b) Stranski–Krastanov and (c) near layer-by-layer growth modes, obtained at different temperatures are shown. STM images (top) of the growth surfaces correlate nicely with the RHEED oscillations (bottom). Adapted from Krishnan (2016).

OUP CORRECTED – FINAL

Transmission High-Energy Electron Diffraction 501 Example 8.4.2: The following figures represent RHEED patterns of the growth of Bi on graphite at (a) room temperature as a function of growth up to 16 MLs of Bi, and (b) after 8 ML growth at two different substrate temperatures. What can you say about the growth of Bi on graphite based on the RHEED data? Please look up Zayed and Esayed-Ali, Phys. Rev. B 72, (2005): 205426, for details. 0.5 ML

0.75 ML

1 ML

002

2 ML

4 ML

8 ML

16 ML

006 004 002

(b) RHEED intensity (arb. units)

(a) Graphite

250 Ts = 300 K Ts = 373 K

(002)

200 (004) 150 Ts = 300 K Ts = 373 K

100

(006) 50 180

200

220

240

260

Row (pixels)

Solution: From (a), we can initially see the spot pattern of graphite. At 0.5 ML we can see the decay of the graphite spot intensity and the appearance of a diffuse background. As the thickness of Bi increases, we can see the appearance of a Bi spot with increasing intensity up to 8 ML. The spots can be indexed as the rhombohedral structure of Bi. There is no intensity oscillation in the intensity of the graphite spot, suggesting growth of three-dimensional islands. However, at 16 ML of Bi elongated RHEED streaks are visible. This suggests coalescence and formation of asymmetric shape crystallites. (b) No change in the relative spot position is observed for the two different substrate temperatures, indicating no change in film growth orientation. However, narrower peaks are observed at higher temperature (373 K) compared to the room temperature (300 K) growth. This suggests that films grown at 373 K have an enhanced crystallite size and/or a higher degree of orientation order, when compared to 300 K growth.

8.5 Transmission High-Energy Electron Diffraction We now describe the diffraction of high-energy (>100 keV) electrons, also referred to as fast electrons to distinguish them from the bound or itinerant electrons in the material, in the context of observations in a TEM. A TEM is

280

300

320

OUP CORRECTED – FINAL

502 Diffraction of Electrons and Neutrons a most versatile instrument (§9.2), routinely used in materials analysis, and which is capable of providing comprehensive diffraction, imaging, and spectroscopic information of thin foil specimens. Moreover, such information can be obtained in a TEM from small volumes, with excellent microstructural sensitivity, and at the highest spatial resolution. The interaction of a fast electron beam with a specimen (see §5.3.2 and §9.3) is complex and includes elastic, inelastic, coherent, and incoherent scattering (§8.5.1), as well as secondary processes such as the emission of element-specific, characteristic X-rays (Fig. 5.3.5). In this section and the next, we focus on elastic scattering and the diffraction of electrons in a TEM, including some typical applications in materials research. Various imaging methods, the associated mechanisms of contrast in a TEM, and the two important analytical methods of energydispersive X-ray spectroscopy (EDXS) and electron energy-loss spectroscopy (EELS), implemented in a TEM, are discussed in §9. Note that a TEM, when it includes EDXS and/or EELS spectrometers, is also called an analytical electron microscope (AEM). After we discuss TEM/AEM in §9, we present details of the complementary technique of scanning electron microscopy (SEM) in §10. We begin with elastic scattering (§5.3.2) as this interaction is fundamental to electron diffraction from crystals as well as the observation of image contrast in a TEM. As discussed in §8.1, fast electrons interact strongly with the atomic nuclei, in addition to the outer electrons, and the resultant Coulomb forces cause their elastic scattering. Typically, the angle of scattering, 2θ , is small (∼1–2◦ ), and in the process of scattering both kinetic energy and angular momentum are conserved. In other words, the fast electrons are scattered elastically without any loss of energy. Note that at high incident energies and large scattering angles, the transfer of energy from the fast electron to the nucleus does take place with multiple consequences, including back-scattering and sputtering (§5.3.2). The specific case of back-scattered electrons, in the context of image contrast in a SEM, is discussed in §10.3.3. Similar to X-rays, the interaction of the fast electrons with a periodic crystal can be characterized by Bragg diffraction (Fig. 8.5.1a) and the associated Laue criterion (§4.3) in reciprocal space (Fig. 8.5.1b). However, unlike X-rays, the wavelengths of fast electrons, λe , is much smaller than the typical inter-planar spacing, dhkl , in crystalline materials (Fig. 4.3.5). As a result, the radius of the Ewald sphere is very large with a very flat surface, and it intersects the reciprocal lattice in multiple zones (Fig. 4.3.4). As for the diffracted intensities on the exit surface of the thin foil (Fig. 8.5.1c) in the simplest column approximation (§8.5.4), it is assumed that the amplitudes, Ag , of the Bragg diffracted waves are small compared to that (A0 ) of the forward-scattered beam. Since the fast electrons are scattered strongly by crystals, this approximation known as the kinematical theory of diffraction is only valid for thin foils, typically less than a few nm in thickness. Simply put, the physical principles of kinematical scattering (§8.5.3) are not very different from that described in terms of the structure factor for X-rays (§7.4), but

OUP CORRECTED – FINAL

Transmission High-Energy Electron Diffraction 503 (a)

Incident Beam

100–300 nm (specimen)

Multiple Scattered beam

Diffracted Beam

(b)

(c)

2θ k

k0

ghkl

Ewald Sphere

Incident Beam

Forward Scattered Beam –g1

0

Exit Diffracted Surface Beam g1

Figure 8.5.1 Diffraction of fast electrons in transmission through a thin foil specimen, illustrating (a) Bragg condition for the incident and diffracted beams in real space, (b) Ewald sphere construction and the Laue criterion in vector form, and (c) the column approximation (see text for details).

with appropriate modifications for (a) electron scattering factors (§8.2), different from those for X-rays, (b) the much smaller wavelength, λe , of the fast electrons, and (c) the shape of the object or thin foil through which it must pass. For most crystals, the thin foil approximation is strictly not valid and the intensity is transferred from the forward-scattered beam to the diffracted beam as the fast electron propagates in the crystal. Further, it is highly probable that after propagating a few more nanometers into the foil, the diffracted beam transfers intensity back to the forward-scattered beam. This back and forth variation in the amplitudes and intensities of the forward-scattered and diffracted beams is referred to as dynamical diffraction. A comprehensive theory of dynamical diffraction of fast electrons requires a quantum mechanical solution of the Schrodinger equation in the periodic potential of the crystal. Further, since the fast electrons experience the periodic potential of the crystal, their wave functions must also reflect the symmetry of the crystals. Such solutions of the Schrodinger equation that reflect the translational symmetry are called Bloch waves (Kittel, 1986) and the theory of electron diffraction in terms of Bloch waves was originally developed by Bethe2 [6]. The dynamical theory of electron diffraction also defines a characteristic distance in the material where the wave amplitude is transferred back and forth between the forward-scattered and diffracted beams. This distance, known as the extinction distance, ξ g , for a specific diffracted beam (or reflection) defined by the reciprocal lattice vector, g, is directly proportional to the volume of the unit cell, Vc , and inversely to the structure factor, F(θ B ), at the exact Bragg angle, θ B , for that reflection as described in (8.5.19). We outline the rudiments of the dynamical theory of electron diffraction in §8.5.4, and describe a method to measure the extinction distance by CBED in §8.6.2.

2 Hans A. Bethe (1906–2005) was a German–American physicist who received the Nobel Prize in Physics in 1967 and was cited for “his contributions to the theory of nuclear reactions, especially his discoveries concerning the energy production in stars.”

OUP CORRECTED – FINAL

504 Diffraction of Electrons and Neutrons

8.5.1 Coherent, Incoherent, Elastic, and Inelastic Scattering Recall our discussion of the propagation of light in transmission through a pin hole (Fig. 6.3.1) and a grating. When the spherical waves from any two openings arrive in phase (the path difference is an integral number of wavelengths) at a specific point on the screen, they constructively interfere, and appear bright (Fig. 6.4.1). At other points on the screen, the waves arrive out of phase, interfere destructively, and disappear. As a result, taking this interference process for the entire grating into consideration, we see alternative bright and dark lines on the screen (Fig. 6.6.7). The diffraction of electrons in transmission through crystals is analogous to the grating; however, the atomic arrangement is in three dimensions and a similar path difference argument for the waves scattered by each atom then predicts a regular “interference” or diffraction pattern of beams/spots instead of lines (see discussion on page 368). In general, we say that the scattering from a set of different atoms located at positions, ri , in a material is coherent if the amplitude, Acoh , of the scattered wave is Acoh =



Ari

(8.5.1)

ri

where Ari is the amplitude of the scattered wave from each ri . The total coherent diffraction intensity measured for waves scattered from all sites is

Icoh

 2 

   = Ari  r 

(8.5.2)

i

depending on the constructive or destructive interference of the different wave amplitudes, Ari . Note that even though it is not explicitly indicated, the intensity depends on the relative phases of the scattered waves and the position, ri , of the scattering atoms. Furthermore, recall the Thomson model of the scattering of an X-ray wave by a bound electron (§7.2.1), where the electric field component of the X-ray wave drives the bound electron into oscillation. This “oscillator” then reradiates at the same frequency, but with a well-defined phase difference (of π for X-rays) between the scattered and incident waves. To reiterate, such scattering with a well-defined phase difference is also termed coherent. Alternatively, the “oscillator” electron, may be coupled to another electron in the solid; further, we can reasonably assume that this coupling in a material, being quantum mechanical, has multiple degrees of freedom. Then, following interaction with the oscillator electron with multiple pathways of excitation, the phase of the scattered wave cannot be easily predicted, and such scattering is called incoherent. Thus, the scattering at all positions, ri , in the material does not preserve any phase relationship between the incident and scattered waves. Then, the total incoherent intensity

OUP CORRECTED – FINAL

Transmission High-Energy Electron Diffraction 505 Iincoh

 2 Ar  = i

(8.5.3)

ri

is the sum of the intensities of the individual waves scattered from each position, ri . Further, the angular dependence of incoherent scattering for N atoms is not different from that of a single atom. Apart from coherent and incoherent scattering, interactions can also be elastic or inelastic (§1.3.1). The former involves no difference in energy between the incident and scattered waves; in contrast, the latter always involves a change in energy. Typically, diffraction experiments involve elastic and coherent scattering, whereas spectroscopy measurements are generally due to inelastic and incoherent scattering.

8.5.2 Basics of Electron Diffraction in a Transmission Electron Microscope The principles of the diffraction of waves by a periodic crystal lattice were introduced in §4.3, formulated both in real space (Bragg’s law, Fig. 4.3.1) and in reciprocal space (Ewald sphere construction, and the Laue criterion, Fig. 4.3.2). In the latter formulation, diffraction maxima are excited when the Ewald sphere passes through a reciprocal lattice point. In §7, we used both real and reciprocal lattice approaches to describe the diffraction of X-rays. It was shown that, if the lattice is not primitive and contains more than one atom in the unit cell, not all points in reciprocal space permitted by the Laue criterion are observed in diffraction. Such systematic absences (§7.6), i.e. which reflections are allowed and which are not, are determined by the structure factor (§7.4). Section 7.5 shows examples of structure factors for many crystal structures calculated for X-ray scattering. The diffraction of electrons in transmission can be understood quantitatively using the same formulism as that used for X-rays, provided we account for some principal differences in scattering between X-rays and fast electrons. First, the wavelength of electrons, λe , for the range of energies encountered in TEM (>100 keV), is very small compared to typical interatomic distances in crystals. Hence, we expect the radius of the Ewald sphere to be very large (Fig. 4.3.4) when compared to the reciprocal lattice unit cell dimensions. As a result, the observed electron diffraction patterns are a planar cross-section of the reciprocal lattice of the crystal. Second, in calculating the structure factors, Fhkl , for diffraction intensities, the atomic scattering factors, fe , for electrons, (8.2.3), corrected for relativistic effects, (8.2.4), must be used. Third, in transmission through thin foils, we are always interested in the intensity of the scattered or diffracted beams at the exit surface of the specimen (Fig. 8.5.1c). Hence, the diffracted intensities are not only dependent on the structure factor (defined by the details of the unit cell) but also on the external shape of the specimen/crystal. The latter, called the shape transform, also modifies

OUP CORRECTED – FINAL

506 Diffraction of Electrons and Neutrons the shape of the reciprocal lattice points–one example discussed is the reciprocal lattice rods (Fig. 8.3.2) in surface diffraction—allowing for diffraction to be observed even when the Ewald sphere does not exactly coincide with a reciprocal lattice point, but passes in the near vicinity of such a point. Fourth, and this bears repeating, since the fast electrons are strongly scattered by atoms, multiple scattering is routine and the intensity of the diffracted and forward-scattered beams dynamically oscillate with thickness. k0 O r β α

8.5.3 Kinematical Theory of Electron Diffraction P

k

t

2θ

Figure 8.5.2 Transmission electron scattering geometry for a thin foil specimen of thickness, t, considering one atom, O, at the origin and the other, P, at a general position, r. The incident, k0 , and diffracted, k, wavevectors are also indicated.

We now calculate the diffraction intensities at the exit surface (Fig. 8.5.1c) for a fast electron beam incident on a crystalline thin foil. We consider two atoms, one at the origin, O, and the other at the general position, P, given by the vector, r, in the crystal (Fig. 8.5.2). For the incident beam, k0 , we are interested in determining the intensity of the diffracted beam, k. The path difference, l, for the two scattered waves emanating from the positions O and P, along the direction, k, is 

l = r cos α − r cos β = r(cos α − cos β) = r

k·r k0 · r − | kr | | k0 r |

(8.5.4)

However, for elastic scattering, |k| = |k0 | = 2π/λ, the phase difference between the two waves is

φ = l

2π = (k − k0 ) · r λ

(8.5.5)

Note that when the Bragg condition is satisfied, the Laue criterion gives k − k0 = ghkl , where ghkl is any reciprocal lattice vector of the crystal. Further, at the Bragg orientation, ghkl · rl = 2mπ , where m is an integer. We use an approach similar to that used to calculate the structure factor, (7.4.7), to calculate the resultant amplitude, A(θ ), of the wave scattered in the direction, k, at an angle, θ (Fig. 8.5.2). It is given by A (θ ) =



fei (θ)exp[i (k − k0 ) · r]

(8.5.6)

i

where fei , (8.2.3), is the electron atomic scattering factor for atom, i. To calculate the amplitude of the diffracted beam at the exit surface, the summation in (8.5.6) has to be carried out over all atoms, i, in the thin foil along the path of the beam. We can simplify the summation by breaking down the coordinates of all the atoms, r, into two parts, i.e. r = ri + rl , where ri , i = 1 . . . n, are the coordinates of all the atoms within the unit cell, and rl , l = 1 . . . N, are the coordinates of all the unit cells in the three-dimensional crystal lattice encountered by the fast electron. Then

OUP CORRECTED – FINAL

Transmission High-Energy Electron Diffraction 507 A (θ) =

n

fei (θ )exp[i (k − k0 ) · ri ]

i=1

N

exp[i (k − k0 ) · rl ] = Fhkl G

(8.5.7)

l=1

The first term, Fhkl , in (8.5.7) is the structure factor discussed earlier for X-ray scattering in §7.4, but is now modified using the appropriate electron scattering factors, fei (θ ), discussed in §8.2. Further, the structure factor is only relevant when the Laue criterion, k − k0 = ghkl , is exactly satisfied. Then, following (7.4.7), at the exact Bragg angle, θ B , it can be calculated as Fhkl =

n

fei (θB ) exp 2πi (hui + kvi + lwi )

(8.5.8)

i=1

To get the total amplitude at the exit surface, the structure factor, Fhkl , is multiplied by the contribution from the lattice. The shape transform or the lattice amplitude, G, in (8.5.7) is determined by the overall shape of the crystal, and includes a summation over all N unit cells in the crystal in the path of the electron beam. We now assume that the intensity, I 0 , of the primary beam remains unchanged as it propagates in the crystal, and we also allow for small deviations from the exact Laue condition, k − k0 = ghkl , to contribute to the diffraction. This deviation is called the excitation error, s (= sx , sy , sz ), and connects the reciprocal lattice vector, ghkl , to the Ewald sphere along the direction of the incident beam (Fig. 8.5.3a). Most importantly, the Laue criterion, (4.3.7), is now modified as (b)

Incident beam, k0 Ewald sphere

Diffracted beam, k θ

2θ

g =

s g+

Origin of reciprocal space

g

s>0 (hkl)

FWHM~1/t

(a)

z 002 202 – 022

222 000

Intensity (a.u.)

111

200 y

s 0, a condition frequently used in imaging defects. Alternatively, if the thin crystalline foil is of uniform thickness, but is bent and not flat, the excitation error, s, now changes spatially with position. If the variation in the excitation error, s, is uniform, it will also produce an intensity distribution with position, and these are known as bend contours (Fig. 8.5.8). The dynamical theory of electron diffraction overcomes the limitations of the kinematical theory, and is discussed in more specialized texts—see Hirsch

2πs

Top of Foil

t

Re

1

sul tan t

Am pli tud e

Bottom of Foil

Figure 8.5.6 Amplitude–phase diagram for the lattice contribution to the intensity in the kinematical theory.

OUP CORRECTED – FINAL

512 Diffraction of Electrons and Neutrons

Specimen 200 000

Diffraction pattern

Equal Thickness Fringes

3μm

Vacuum

Figure 8.5.7 Bright field image of a AlCu alloy with the [200] reflections excited, showing thickness fringes. Courtesy JEOL Ltd.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

e d c a b

f 10μm

Figure 8.5.8 Bend contours observed in thin foil of mica. (a) Bright field image. (b)–(f) Dark field images formed by reflections indicated in (g) the selected area diffraction pattern from the same area shown in (a). Notice that in this bent crystal, the image shows a strong intensity in a dark field image at positions where the Bragg condition is satisfied. Courtesy JEOL Ltd.

et al. (1965), Cowley (1984), Williams and Carter (1990), Reimer (1993), and Fultz and Howe (2013). Here, we limit ourselves to a very brief mention of the assumptions made in its formulation and present one of its key results. Unlike the kinematical theory, in the dynamical theory it is now assumed that as the electron propagates in the crystals, each thickness element scatters its intensity back and

OUP CORRECTED – FINAL

Transmission High-Energy Electron Diffraction 513 forth between the forward-scattered and diffracted beams. A system of coupled equations (known as the Howie–Whelan equations, [8]) describes the amplitudes of the diffracted, Ag (z), and forward-scattered, A0 (z), beams, as they propagate through the thickness, z, in the thin foil specimen, and when they are solved—as in the dynamical theory—correctly predict the intensities, I g , of the diffracted beam. The solution, similar in form to (8.5.16), is given by sin2 π t s¯

Ig ∼

(π s¯)2

(8.5.17)

where, the excitation error, s, is modified as 

1 s¯ = s + 2 ξg

1/2

2

(8.5.18)

and includes the extinction distance, ξg , given by ξg ≈

π Vc λe Fg (θB )

(8.5.19)

Here, Vc is the volume of the unit cell, Fg (θB ) is the structure factor for the specific reflection, and λe is the electron wavelength. ξg is the difference between two thickness extinctions, and as mentioned before is a characteristic distance in the material where the wave amplitude is transferred back and forth between the forward-scattered and diffracted beams. For TEM electrons of energies ∼100–200 keV, and typical low-order reflections, the extinction distance, ξg ∼ 20 − 200 nm. So far, we have discussed perfect crystals, but most materials are imperfect and harbor various defects such as dislocations (§4.1.9.2) and stacking faults (§4.1.9.3). The analysis of defects, particularly dislocations, using diffraction contrast is discussed further in §9.5.2.

Example 8.5.1: Calculate the extinction distance for Cu111 for electron incidence at 100 keV. Solution: The extinction distance is given by (8.5.19). Cu is FCC with a0 = 0.361 nm. Thus, the volume of its unit cell, Vc = (0.361)3 = 0.047 nm3 . The wavelength of 100 keV electrons is λ = 0.0037 nm. The interplanar spacing d 111 = 0.208 nm (see Table 4.1.2 and Example 7.5.1) From Example 8.2.1, fe (100 keV) = 0.3448 nm. From Example 7.5.1, the structure factor F 111 = 4fe = 4 × 0.3448 = 1.3792 nm

OUP CORRECTED – FINAL

514 Diffraction of Electrons and Neutrons

Applying

(8.5.19),

we

get

an

extinction

distance,

c ξg ≈ λ πV = F (θ ) e g

B

= 28.9 nm. Note that the extinction distance depends on the lattice parameter (through the unit cell volume, Vc ), the atomic number (through the structure factor, F g ), and the acceleration voltage (through the electron wavelength, λ). π x 0.047 0.0037 x 1.3792

8.6 Transmission Electron Diffraction Methods There are three principal types of electron diffraction patterns produced in a TEM: (a) SAD with parallel illumination, producing spot (single crystal grains) or ring (polycrystalline materials) patterns, (b) Kikuchi line patterns involving both inelastic and elastic scattering, particularly from thicker specimens, and (c) CBED patterns obtained from small areas of the specimen using a focused incident probe. These are discussed in the next three sections.

8.6.1 Selected Area Diffraction: Ring and Spot Patterns For crystalline specimens, TED patterns simply correspond to a planar section of the reciprocal lattice (Fig. 4.3.4; Fig. 8.5.3) normal to the incident beam direction and recorded in the form of an image. Moreover, in a TEM the diffraction pattern is obtained from the area of interest in the specimen by using a selected area aperture of diameter, d SA , placed in the first image plane of the objective lens (Fig. 9.2.9). The diffraction pattern for crystalline specimens is a magnified image of the reciprocal lattice array/spots on the surface of the Ewald sphere (Fig. 8.6.1). In the observed diffraction pattern, the measured distance, Rhkl , between the forward scattered beam (origin) and any diffracted beam (spot), can be related to the appropriate lattice spacing, dhkl , for the specific reflection, ghkl , in terms of the camera length, L, as 3 Modern TEMs with digital cameras will often include/embed the camera constant, λL, when recording diffraction patterns. Moreover, intermediate lens astigmatism should be adjusted for round ring patterns, projector lens distortion affects outer rings, and therefore unless you have done the calibration yourself this should not be taken at face value.

λe L = Rhkl dhkl

(8.6.1)

As a result, to obtain accurate measurements of dhkl , it is necessary to know the value of the camera length, L, which varies according to lens setting and does not correspond at all to the physical specimen-camera distance, for a given instrument (TEM) under specific experimental conditions of specimen position (height, z), objective lens excitation current etc.3

OUP CORRECTED – FINAL

Transmission Electron Diffraction Methods 515

Thin Foil

Camera Length, L

2π λe

k0

k Ewald Sphere 2θB ghkl

Rhkl

(a)

(b)

Reciprocal Lattice

Observed Diffraction Pattern

Figure 8.6.1 A schematic diagram of the formation of an electron diffraction pattern in transmission, with a camera length, L, that is calibrated for specific settings (lens excitations) in a TEM.

(c)

311 440 220 400 511

Figure 8.6.2 (a) A Debye–Scherrer type ring pattern obtained from a small-grain, polycrystalline copper specimen that can be used to determine the camera length, L. (b) Image of monodisperse iron oxide nanoparticles, ∼25 nm in diameter, dispersed on an ultrathin carbon film. (c) Ring pattern from (b) that can be indexed as magnetite (Fe3 O4 ). Courtesy Dr. Ryan Hufschmid.

In practice, for a specific TEM, the cameral length is best determined experimentally by measuring the Debye-Scherrer ring pattern (§7.9.3) from a standard, small-grain, polycrystalline material, such as an evaporated gold thin film (Fig. 8.6.2a). Here, all the radii, Rhkl , for the rings are measured accurately from the observed pattern, all inter-planar spacing, dhkl , are known, the electron

OUP CORRECTED – FINAL

516 Diffraction of Electrons and Neutrons wavelength, λe , is calculated from the acceleration voltage, and then, L is determined accurately from (8.6.1). For crystalline materials, the electron diffraction patterns for incident beam directions along low-index zone axis orientations (Fig. 4.1.10) reflect the crystallographic symmetry of its reciprocal lattice and as such, display highly symmetric patterns. Figure 8.6.3 shows examples of spot patterns for single crystals with FCC, BCC, diamond cubic, and HCP structures. Example 8.6.1: Index the ring pattern in Figure 8.6.2a and determine the camera length, L, of the instrument. The manufacturer claims that L = 65 cm, for this setting. What is the error, if any? Assume 100 kV incident electrons. Solution: The wavelength for 100 keV electrons, λ = 0.037 Å, and aCu = 3.61 Å. This is the classic FCC pattern following the sequence, two-together, one alone, two-together . . . , and the first five rings can be indexed as 111, 200, 220, 311, 222 . . . . We now solve for the camera length, L, using (8.6.1), in the form of a table: Reflection Ring diameter measured (cm) Ring radius, r (cm) dhkl (Å) λL = r dhkl (Å cm) Camera length, L (cm)

111 2.22 1.11 2.08 2.31 62.43

200 2.58 1.29 1.81 2.33 62.97

220 3.65 1.83 1.28 2.34 63.24

311 4.28 2.14 1.09 2.33 62.97

222 4.56 2.28 1.04 2.37 64.05

Average value of the camera length, Lavg = 63.35 cm. The error in the manufacturer’s value is ∼2.5%.

A characteristic feature of TED, because multiple scattering is prevalent, is double diffraction. As a consequence, a diffracted beam, g1 , can be re-diffracted by another reflection, g2 , either in the same or another crystal/grain, to produce an observable double-diffracted beam, g1 + g2 , which may or may not be allowed for the crystal based on its structure factor. Such reflections, also indicated in Figure 8.6.3 for various zone axis orientations, display intensities that are not readily interpretable (see Example 8.6.1). Thus, specific reflections that are forbidden by the extinction rules in structure factor calculations, §7.5, are nevertheless observed in electron diffraction. Finally, to analyze the crystal structure of an unknown specimen/material by electron diffraction, it is important to obtain sections of its reciprocal lattice in multiple directions of electron incidence. This is best done by tilting the crystal, for a fixed electron beam incidence direction, through a large angular range (±45–60◦ ), depending on the goniometer included as part of the specimen stage in the TEM) around any one crystallographic axis. Then, the diffraction patterns

OUP CORRECTED – FINAL

Transmission Electron Diffraction Methods 517 (a)

45°

45°

54.74°

60

54.74° 54.74 4°

60 – 220

– –– 220 200

45°

– 020

45°

35.26° 35.

35.26°

200 220

– – 202

–– 220

60

– – –– A 111 B 111 – –– –– 022 022 000 – –– 111 111 200

A 020 B 000

– 220

– 200

A – 022

A 220

202 2 A = = 1.155 B = z = [011] B √ √3 3

A √2 = = 1.414 B = z = [001] B 1

–– 022

A

B = z = [111]

(b) 35.26°

45° – 200 –– – 110 110 A B – 020 000 – 110 110

45°

54.74°

45° 020

60

–– – – – 211 200 211 –– 211 B A C –– – 011 000 011

60 35.26°

60

– 112

– –– 211 200 211

200

√4 4 A √ = = 1.414 B = z = [001] B √ √2 2

(c)

2 B A √6 001] = = 1.414 B = z = [001] = = 1.732 C √2 B √2

B = z = [111]

60°

45°

45°

– 011 – 121 00 000 110 101

A AA

54 4.74°° 54.74°

54.7 74° 54.74° 60°

45° A 000

45°

– 220 B

60°

35.26° 5.26

A

– 000 – 022 111 200

040

– 022

A

A

220 400 440 B 4 = = 1.414 B = z = [001] A √8

35.2 6° 35.26°

– – 111

B

000 – 202

2 B = = 1.155 B = z = [011] A √ √3 3

A – 220

B = z = [111]

(d) 61.38°

61.38°

57.79°

00.2 28.62° 8.62

C

01.1

B A 00.0 01.0 0 –– 01.1 – –– 01.2 00.2

28.62° 28.62°

– C B = 1.09 = 1.139 B = z = [21.0] A A

32.21°

00.2

57.79° – 21.2

A

B

00.0

C = 1.587 A

C

30° 30°– 30° 30°° – 12.0 11.0 01.0 32.21°

– 21.2

B = 1.876 B = z = [01.0] A

30° 30°

30 0° 30°

A A A 00.0 10.0

30°°

B = z = [00.1]

Figure 8.6.3 Single crystal spot transmission electron diffraction patterns for (a) FCC, (b) BCC, (c) Diamond cubic, and (d) HCP structures. Unfilled diamonds show reflections that appear due to double diffraction. The beam direction, B, and the zone axis, z, defined in §4.1.4, satisfying the crossproduct rule, (4.1.8), are also indicated.

OUP CORRECTED – FINAL

518 Diffraction of Electrons and Neutrons (a)

(b) 020

(c)

002 200

200

002 110

Figure 8.6.4 SAD patterns of Fe14 Nd2 B along three different orientations. The halo around the forward-scattered (transmitted) and diffracted spots, as well as the faint background intensity observable in (b) are attributed to the presence of inelastically and incoherently scattered electrons. Adapted from Shindo and Hiraga (1998).

are recorded in multiple zone axis orientations. Figure 8.6.4 shows an example of electron diffraction of a permanent magnet alloy, Nd2 Fe14 B, taken along three different zone axes orientations. Note that in Figure 8.6.4b, reflections of the type h + l = 2n + 1 are not observed; however, in both (a) and (c) they are present due to double diffraction. Tilting of the crystals, from one zone axis orientation to another, where possible, is accomplished by following directions indicated by Kikuchi lines/patterns. The Kikuchi lines originate from a superposition of inelastic and elastic scattering of electrons in thicker crystals, and are discussed in the next section. Example 8.6.2: Index the diffraction pattern of a specimen of silicon shown in the following image. The diameter of the spots corresponds to their intensity. Are the weak reflections, enclosed in the dotted circle, allowed from the structure factor for this structure? Then, explain their presence in the diffraction pattern.

15mm ~55° 13mm

Solution: Silicon is a diamond cubic crystal structure. Comparing with Figure 8.6.3, we can easily index this as a B = z = [011] pattern and assign the reflection as in the following figure.

OUP CORRECTED – FINAL

Transmission Electron Diffraction Methods 519

200 111 022

111 022

000 111

111 200

From Problem 7. 1, i.e. structure factor calculations for the diamond cubic structure, we know that Ihkl = 0, when h + k + l = 4n + 2, where n is an integer. Thus, the 200 reflections are forbidden for Si, and should be absent. However, the 111-beam with structure factor, F 111 = 0, can act as a new incident beam and be re-diffracted by the (111) planes. In other words, we can have double or dynamical diffraction, combining the two reflections,   i.e. 111 + 111 = (200), to faintly excite the 200 reflection. Thus, the 200 reflection is only “kinematically forbidden”, and may be excited by dynamical diffraction. Unlike for X-rays, where single scattering dominates, double diffraction is a common feature in electron diffraction.

8.6.2 Kikuchi Lines, Maps, and Patterns In crystalline thin foils that are reasonably thick, i.e. about half the penetration distance of the fast electron, and with low defect densities, the background intensity between the Bragg spots in TED displays bands of bright and dark lines. Occurring in pairs, these parallel bright and dark lines known as Kikuchi lines, named after their discoverer [9], are used to determine the sign and magnitude of the excitation error, s. They also define the sense of tilt of the specimen, reveal true crystal symmetry unlike the SAD patterns, and guide the experienced TEM operator in tilting a crystalline specimen from one zone axis orientation to another. The underlying mechanism and geometry of the formation of Kikuchi lines can be understood from the diffraction of those electrons that have previously undergone inelastic scattering (Fig. 8.6.5a). A thin crystal is oriented such that the incident beam does not satisfy the Bragg law of diffraction for the planes (hkl), as shown in the figure. We focus our attention on the inelastically scattered electrons at any point, O, inside the crystal. Figure 8.6.5a shows the distribution of the inelastic electrons is forward peaked; moreover, we assume that the energy losses ( 0, the excess (bright) Kikuchi line is further away from g, at a distance, x, as shown, and for s < 0, it is closer to the transmitted beam, O, than g. If the camera length, L, is known, and the distances Rhkl and x (Fig. 8.6.6) are measured, from the scattering geometry we can show that s∼

xλe 2 g Rhkl

(8.6.2)

In addition, for contrast analysis not only is the excitation error, s, an important parameter, but it is important to tilt the crystal such that a specific order of the reflection (g, 2g, 3g etc.) is excited or is in the exact Bragg orientation. This is also accomplished by using Kikuchi lines (Fig. 8.6.7). Last, but not least, at the exact zone axis orientation, the Kikuchi lines can be drawn as perpendicular bisectors for every allowed reflection, ghkl . Figure 8.6.8a shows an example of the Kikuchi line construction for the [001] zone axis for an FCC crystal. The Kikuchi pattern from one zone axis can be extended to a second zone axis (pole) by following a pair of Kikuchi lines that are common to both poles (Fig. 8.6.8b). In this manner a complete Kikuchi map of the angular relationship between different poles, useful for “navigation” in reciprocal space while performing diffraction experiments in a TEM for any particular crystal

OUP CORRECTED – FINAL

522 Diffraction of Electrons and Neutrons

Specimen

2θB

k0

θ kg

s=0 Figure 8.6.6 The Ewald sphere construction showing two different conditions of the excitation error, s = 0, and s > 0. In practice, this is achieved by tilting the crystal and monitoring the relative orientation of the Kikuchi lines with respect to the Bragg diffraction peaks, as shown on the right.

Ewald Sphere

s

s>0 Rhkl

Diffraction Pattern s=0

Figure 8.6.7 Illustrating the use of Kikuchi lines for tilting to an exact Bragg orientation; from left to right, conditions for exciting ghkl , 2ghkl , and 3ghkl , respectively. Adapted from Williams, Pelton, and Gronsky (1991).

g

x s>0

s=0

s>0

3h3k3l

2h2k2l 000

hkl

000

000

structure can be generated. Figure 8.6.8c shows an example of a Kikuchi map, for the FCC structure. Needless to say, similar Kikuchi maps for other crystal structures, such as BCC, HCP, etc., can be constructed or found in the literature; see, for example, Edington (1975), Thomas and Goringe (1980), and Williams and Carter (1990).

OUP CORRECTED – FINAL

Transmission Electron Diffraction Methods 523 (a)

g020

(b)

– ±111 Kikuchi Lines

–– ±111 Kikuchi Lines

(c)

– 020 [100] 020 [101]

– 200

200

45° Figure 8.6.8 Construction of Kikuchi lines illustrated for the [001] zone axis, by drawing (a) the perpendicular bisectors for all allowed reflections. (b) By extending pairs of Kikuchi lines common to two poles with welldefined crystallographic orientations, a Kikuchi map such as the one for (c) FCC crystals can be established. Adapted from Edington (1975).

8.6.3 Convergent Beam Electron Diffraction (CBED) SAD in a TEM is implemented using an aperture in the first image plane (§9.2.7.1) in front of the diffraction or intermediate lens. However, the spherical aberration of the electromagnetic lens (see §9.2.2 and §9.2.9 for a detailed discussion of lens aberrations) limits the area selected for diffraction, using parallel illumination, to ∼1 μm in diameter. To obtain diffraction from smaller areas of the order of 10 nm, a focused probe with a convergence semi-angle, α (Fig. 8.6.9) is used. If the convergence semi-angle is very small, α > θ B , we

OUP CORRECTED – FINAL

524 Diffraction of Electrons and Neutrons (c)

(b)

(a)

2α

2α Thin Specimen

2α Thin Specimen

Thin Specimen

2θB 2θB

2θB

Spot Pattern

CBED Pattern

Kossel Pattern

α > θB

Figure 8.6.9 Effect of increasing beam convergence semi-angle, α, as compared to the Bragg angle, θ B , on transmission electron diffraction patterns using a stationary probe, illustrating the formation of (a) micro diffraction pattern, (b) convergent beam electron diffraction, and (c) Kossel patterns. observe a Kossel pattern of overlapping discs (Fig. 8.6.10d) with dark and bright line features similar to Kikuchi patterns. We shall restrict our discussion here to CBED patterns. Kossel patterns are discussed in Reimer (1993). A convergent probe includes a large range of incident wave vectors, k0 , within the illuminated cone (Fig. 8.6.9b) and therefore each point in a CBED disc corresponds to a particular incident wave vector. The intensity distribution within the forward-scattered disc and any diffraction disc represent the intensity variations with the excitation error, s. Further, the intensity variation in a disc at the exact Bragg orientation (Fig. 8.6.11) reflects the variation in intensity with the modified excitation error, s¯, defined in (8.5.18), and as plotted in Figure 8.5.3a. A simple method has been developed to utilize the thickness fringes in CBED patterns (Fig. 8.6.11) to not only measure the thickness, t, of the specimen, but also the extinction distance, ξ g , of the material. Each of the minima in the Bragg diffraction disc has a characteristic angular deviation, θ i , that can be measured. The corresponding excitation error, si , can be calculated from the relationship si =

λe 2 dhkl



Δθi 2θB

 (8.6.3)

where all variables are known, or measured from Figure 8.6.11a. Further, the deviation parameter, si , is related to the extinction distance, ξ g , simply as

OUP CORRECTED – FINAL

Transmission Electron Diffraction Methods 525 (a)

(b) 040

220 000

(c)

400

(d)

Figure 8.6.10 Diffraction patterns of a single crystal spinel, MgAl2 O4 , in the [001] zone axis orientation. (a) Selected area diffraction pattern obtained with parallel illumination (fixed k0 ); Notice the absence of the (200) type reflections, forbidden by the structure factor. (b) A convergent beam electron diffraction (CBED) pattern of the zero-order Laue zone (ZOLZ) showing dynamical contrast within the individual discs, including in the forbidden (200) reflections; in this case a medium convergence angle for the incident beam representative of a range of incident vectors, k0 , is employed. (c) The entire CBED patterns showing both the ZOLZ and the first-order Laue zone (FOLZ) with individual disc features resolved. (d) A larger convergence angle results in the overlap of all the individual discs in the CBED pattern, but the Kikuchi lines remain visible and the pattern retains its overall symmetry—these are also known as Kossel patterns.

OUP CORRECTED – FINAL

526 Diffraction of Electrons and Neutrons (b)

θi

(a)

intercept = 12 t si ni

2

slope =

2θB

1 ni

1 ξ2g

2

Figure 8.6.11 (a) Observation of thickness fringes in a convergent beam pattern at the exact (s = 0) Bragg orientation. (b) The fringe spacing can be measured and plotted to obtain both the thickness (intercept) and the extinction distance (slope) for that specific reflection. Courtesy Chuck Echer.

s2i n2i

=−

1 1 1 + 2 ξg2 n2i t

(8.6.4)

where ni is an integer. To determine t and ξ g , it is common practice to plot (si /ni )2 versus (1/ni )2 , assigning, n1 = 1, n2 = 2, etc. If the result is a straight line (Fig. 8.6.11b), then its slope is –(1/ξ g )2 , and its intercept is (1/t)2 . If the result is not a straight line, the numbers, ni , are reassigned different integer values, until a straight-line fit is obtained, and the slope and intercept is then determined to give values of the extinction distance and thickness, respectively. As mentioned, if the diffraction pattern is recorded with a large collection angle along a low-index zone axis (Fig. 8.6.10c), higher order Laue zones, in the form of rings, are observed. Because of the large Bragg angles associated with the HOLZ reflections, they do not appear as discs/spots, but as lines. In addition, every bright line in a HOLZ ring has a corresponding dark line in the forward scattered beam (Fig. 8.6.10b), and can thus be indexed. These HOLZ lines in the forward scattered disc are sharp and appear, unlike Kikuchi lines, even in thin specimens. Moreover, since the HOLZ lines are both sharp and correspond to large Bragg angles, they are very sensitive to small changes in lattice parameters. Further, if the CBED pattern is carefully aligned along the exact zone axis (Fig. 8.6.10b), the symmetry of the pattern reflects the point group symmetry of the crystal. In additions, symmetries present in the forbidden reflections, such as in the (200) reflections in Figure 8.6.10b, is consistent with the space group of the specimen. CBED patterns are used to make elaborate analysis of the symmetry of the crystal at high spatial resolution that is defined by the size of the convergent probe incident on the specimen. Details of symmetry analysis by CBED can be found in [10].

OUP CORRECTED – FINAL

Examples of Transmission Electron Diffraction of Materials 527

8.7 Examples of Transmission Electron Diffraction of Materials There are numerous applications of TED in materials characterization and analysis. Here we present select practical examples, mostly from nanoparticles and thin film heterostructures, to illustrate some basic applications; many more applications in metallurgy and ceramics are discussed in Thomas and Goringe (1980), Williams and Carter (1990), and Fultz and Howe (2013).

8.7.1 Indexing a Single Crystal Diffraction Pattern Many applications require the indexing, or assigning specific reflections, ghkl , to the observed peaks in the intensities of the diffraction patterns. For a single crystal or large grain of the material, it is best done by tilting the crystal to a low-index zone axis orientation and recording the diffraction pattern (Fig. 8.6.4; Fig. 8.6.10a). Then, using the known camera length, L, and wavelength, λe , the distance, Ri , measured for each of the spots can be associated with a specific interdj i planar spacing, di , using (8.6.1). Further, the ratios of any two values R Rj = di , gives the ratios of possible inter-planar spacing. After this, using the measured set of ratios, dj /di , and the symmetry of the zone axis pattern, and comparing them to either a table of allowed ratios or zone axis patterns (Fig. 8.6.3), one can make a first attempt at indexing the reflection as well as the zone axis orientation. The veracity of the indexing can be further cross-checked by using pairs of reflection and applying the zonal equation, (4.1.8), to them (Fig. 4.1.10). For example, Figure 8.6.10 is a [001] zone axis, and the three reflections indexed as (220), (400) and (040), are consistent with the cross-product of the zonal equation, (4.1.8). Methods for indexing more complicated diffraction patterns, often encountered in materials analysis, are discussed in all three specialized books mentioned earlier.

8.7.2 Polycrystalline Materials and Nanoparticle Arrays TED from polycrystalline materials, including an assembly of randomly oriented nanoparticles, can be interpreted in a manner similar to X-ray diffraction from powder specimens (§7.9.2). If the specimen is truly polycrystalline and randomly oriented, its reciprocal lattice is a set of nested spheres, with each of its radii corresponding to a reciprocal lattice vector of the crystal. The intersection of these nested spheres with the relatively “flat” Ewald sphere in electron diffraction produces a set of rings in the SAD pattern (Fig. 8.7.1). The width of each ring, similar to a powder pattern in X-ray diffraction, gives an estimate of the average size of the grains or nanoparticles. Further, the intensity of each ring can be “circularly integrated” to produce a plot (Fig. 8.7.1g) of the intensity as a function of inverse lattice spacing, 1/dhkl , and then compared to the powder diffraction files (§7.11.3) to help identify the crystal structure.

OUP CORRECTED – FINAL

528 Diffraction of Electrons and Neutrons (a)

(b)

(c)

20 nm

20 nm

20 nm

(d)

FeO FeO (220) (311)

Fe3O4 (311)

(e)

(f)

Fe3O4 (311)

2

nm–1

FeO (111)

FeO (200)

(311) 2 nm–1 (111) (200) (220)

2 nm–1

(g)

Magnetite Wüstite

0

2

(111) (220) (311) (400) (422) (511) (400)

4 nm–1

6

8

Figure 8.7.1 Time evolution of the crystal structure and phase purity of iron oxide nanoparticles synthesized by chemical routes in an organic solvent. At some intermediate stage (a, d) the nanoparticles are mainly FeO with a trace of Fe3 O4 . At the end of the synthesis (b, e) the nanoparticles are faceted, uniform in size, but uniformly maghemite (Fe3 O4 ). Circular integration of the ring pattern produces a powder pattern (g) that matches very well with the PDF files for maghemite. However, (c, f ), when a controlled oxidation step is introduced in the synthesis the nanoparticles have the most desirable phase, magnetite (Fe3 O4 ). Adapted from [11].

OUP CORRECTED – FINAL

Examples of Transmission Electron Diffraction of Materials 529 Figure 8.7.1 illustrates these ideas when applied to the synthesis of iron oxide nanoparticles by the decomposition of chemical precursors in organic solvents [11]. Figure 8.7.1a,d shows the particles at some intermediate stage of the synthesis; even though a faint ring indicates the presence of trace quantities of Fe3 O4 (magnetite), it is mainly FeO. At the end of the synthesis, without an oxidation step (Fig. 8.7.1b,e,g) the nanoparticles are predominantly maghemite (Fe2 O3 ), with the integrated intensity (Fig. 8.7.1g) matching very well with its powder diffraction file (PDF) data. Finally, when a controlled oxidation step is included, the nanoparticles attain their equilibrium shape and the SAD pattern is consistent with magnetite (Fe3 O4 ) as confirmed by the indexed rings in Figure 8.7.1f.

8.7.3 Orientation Relationships Between Crystals or Phases Another important application of electron diffraction in materials analysis is to determine the orientation relationship between two or more different crystals in a specimen, such as a precipitate distributed in a matrix, a thin film grown on a substrate, etc. A complete analysis requires the determination of the relationships of both crystallographic planes and directions in the two crystals. Figure 8.7.2a shows the complex microstructure observed in Sm–Co permanent magnet alloys, doped with small amounts of Fe and Zr. It consists of three phases, hexagonal SmCo5 , rhombohedral Sm2 Co17 , and a thin platelet (Z) phase containing most of the Zr additions. The typical microstructure shown includes a network of fine (∼200 nm in diameter) cells of the rhombohedral Sm2 (Co,Fe)17 , separated by cell walls (∼5–20 nm thick) of a Sm(Co,Cu,Zr)5 boundary phase, and a third platelet phase, containing Zr, running across many cells and cell boundaries. A SAD pattern from this microstructure is shown in Figure 8.7.2b, and it is indexed in Figure 8.7.2c. From the diffraction pattern, we can infer the orientation relationship between the phases, and conclude that all the phases are crystallographically coherent. This is important as these alloys are called pinningtype permanent magnets and their magnetic hardness arises from their ability to pin the motion of domain walls in a structurally coherent microstructure. For further details on the role of microstructure in determining the behavior of hard magnets, see Krishnan (2016).

8.7.4 Chemical Order in Materials Here we illustrate studies of chemical order with the simple case of a bilayer thin film of Fe and MnPd, epitaxially grown on a single crystal MgO[001] substrate. MnPd is a chemically ordered L10 structure (you can visualize this as a variant of the CuAu structure; Fig. 7.11.2b) based on a face-centered tetragonal unit cell, and with lattice parameters (Fig. 8.7.3a). It can be grown epitaxially [12, 13], either

OUP CORRECTED – FINAL

530 Diffraction of Electrons and Neutrons

c

(b) [0001]1:5 ≡ [0003]2:17 [0000] [1011]2:17

[2022]2:17 [2111]1:5 ≡ [3033]2:17

0.1 μm

(a)

(c)

[2110]1:5 ≡ [3030]2:17

Figure 8.7.2 Orientation relationship between the different crystallographic phases, SmCo5 and Sm2 Co17 , in the complex microstructure observed in Sm–Co permanent magnet alloys. Courtesy of Dr. L. Rabenberg.

with the c-axis normal (Fig. 8.7.3a) or with the a-axis normal (c-axis in plane) to the film plane (Fig. 8.7.3c). When chemically ordered as mentioned, the Mn and Pd atoms arrange themselves in alternating layers and the inter-planar period doubles. Then, in a typical diffraction pattern (Fig. 8.7.3f), both the regular (002) reflections and the chemically ordered (001) reflections are visible; moreover, the relative intensity of the (001) reflection, compared to the (002) reflection, is a good indicator of the order parameter (§7.11.4), or the degree of chemical order. For a film grown with the c-axis normal, we can clearly see the chemical order, or alternating layers of Mn and Pd atoms perpendicular to the film normal, in the X-ray scan (Fig. 8.7.3b). However, it is not possible to say if an a-axis normal film is chemically ordered or not from an X-ray scan (Fig. 8.7.3d), as the scattering vector, k – k0 , is parallel to the film normal in a symmetric θ -2θ X-ray scan, but the ordering is in planes normal to the film surface (Fig. 8.7.3c). However, a TED pattern from a plan-view specimen with an in-plane scattering vector can register the chemical order in the a-axis normal film (see Figure 4.3.5 and related discussion). This is illustrated for chemically disordered (Fig. 8.7.3e) and ordered (Fig. 8.7.3f ) films, where the latter clearly shows the (001) reflections representing

OUP CORRECTED – FINAL

Examples of Transmission Electron Diffraction of Materials 531 (a)

(c)

(e) a-axis normal

c-axis normal

4.07Å

3.58Å

022 4.07Å

Mn

4.07Å

002

3.58Å

Pd

4.07Å

(b)

(f)

(d) (002) MgO PdMn(001)

(002) MgO PdMn(200)

PdMn(002)

Fe(002)

001

Fe(002)

002 10 20 30 40 50 60 70 80 2θ(deg)

10 20 30 40 50 60 70 80 2θ(deg)

Figure 8.7.3 MnPd is a L10 ordered tetragonal structure that can be grown epitaxially either with (a) the c-axis normal, or (b) the a-axis normal to the film plane. In X-ray θ–2θ scans, the diffraction vector is normal to the film plane and thus it is only able to resolve chemical order (b) in the c-axis oriented film, with the observation of the MnPd (001) peak, but not in (d) the a-axis normal films. However, plan view electron diffraction of as grown ( e) and annealed ( f) specimens clearly show the chemical order in the a-axis normal films. Adapted from [12, 13].

the chemical order, with their fourfold symmetry corresponding to the epitaxial growth of the layers on a (100) surface. Figure 4.3.5 introduced the difference in scattering vectors conceptually between θ -2θ X-ray scans and plan-view electron diffraction. However, Figure 8.7.3 is a practical example illustrating the importance of selecting the appropriate technique with the right scattering geometry for a specific problem.

8.7.5 Diffraction from Long-Period Multilayers We now consider an example of diffraction from a multilayer, with an artificial period of ∼6.7 nm, and consisting of alternate layers of Mo (5.4 nm) and Si (1.3 nm), as clearly visible in the image of a cross-section specimen (Fig. 8.7.4a). The growth of individual layers, particularly Mo, is columnar with some texture. This is reflected in the angular spread of the SAD spot, as shown for Mo (011)

OUP CORRECTED – FINAL

532 Diffraction of Electrons and Neutrons Figure 8.7.4 A cross-section specimen of a [Mo5.4 Si1.3 ]n multilayer, grown on a single crystal Si substrate. (a) A bright field (BF) image, clearly showing the period of the bilayers, and the columnar growth of each of the layers. (b) The selected area diffraction pattern from the multilayer structure; the central region is blown up, and the fine spots indicate a periodicity of ∼6.7 nm corresponding to the periodicity of the Mo–Si bilayers observed in the BF image. How do we know that the two layers are Mo and Si? And, what is the atomic structure and is there intermixing of Mo and Si at the interface? These questions are answered by high spatial resolution spectroscopy and imaging methods in a TEM (Fig. 9.5.9) and discussed further in §9.5.3.

(a)

20nm Si substrate

(b)

Mo 011

Mo ~ 5.4nm Si ~1.3 nm

in Figure 8.7.4b. Further, the fine spacing of the superlattice reflections, shown in the magnified image of the central portion of the diffraction pattern, corresponds to the observed periodicity of ∼6.7 nm. Note that this analysis is straightforward as the Mo (011) reflection serves as an internal calibration, without knowing the cameral length, to measure the multilayer periodicity.

8.7.6 Twinning A twin is a common phenomenon observed in materials and is important in determining their mechanical properties. Typically, a twin can be formed by shear, such that all atoms on one side of the twin boundary (or plane) appear to be in positions of mirror image with respect to the other (Fig. 8.7.5a). The twin planes depend of the crystal structure, and are {111} and {112} for FCC and BCC crystals, respectively. Alternatively, because the atoms are spherically symmetric, a twin can also be considered as arising from a 180◦ rotation about an axis normal to the twin plane. Since the reciprocal lattice points are simply related to the real lattice (§4.2), the reciprocal lattice for the twinned crystal defined by the indices (PQR), is related to the original crystal, defined by the indices, (pqr), by a simple

OUP CORRECTED – FINAL

Examples of Transmission Electron Diffraction of Materials 533 Twin Plane

(a) Matrix Crystal

(b) Twinned Crystal

Matrix A B C A B A C B Stacking Sequence

Figure 8.7.5 Twinning in an FCC crystal, showing (a) the position of atoms for a (110) section of a twin  on 111 ; notice the reversal of the ABC stacking sequence of the matrix in the twin to ACB. (b) The diffraction pattern (indexed) observed for such twinning in an FCC crystal in the [110] zone axis pattern.

Twin

A

Adapted from Edington (1975).

transformation matrix, Thkl , defined by rotation of 180◦ about the normal to the twin plane (hkl), as (PQR) = Thkl (pqr)

(8.7.1)

where the twinning matrix, Thkl , depends on the crystal structure. For a cubic crystal, it is given by ⎛

Thkl

h2 − k2 − l 2 1 ⎜ ⎝ =  2 2hk h + k2 + l 2 2hl

2hk −h2 + k2 − l 2 2kl

⎞ 2hl ⎟ 2kl ⎠ −h2 − k2 + l 2 (8.7.2)

As mentioned, for an FCC crystal twinning occurs on the {111} plane. Using these values for hkl in (8.7.2), we get for the transformation matrix:

TFCC 111

⎛ −1 1 ⎜ = ⎝ 2 (3) 2

2 −1 2

⎞ 2 ⎟ 2 ⎠ −1

(8.7.3)

Figure 8.7.5b shows a diffraction pattern for an FCC crystal structure observed  for the [110] zone axis, but twinned on 111 ; it is left as an exercise to the reader to show that this pattern is consistent with (8.7.3).

OUP CORRECTED – FINAL

534 Diffraction of Electrons and Neutrons Example 8.7.1: The following figure is an electron diffraction pattern of a steel microstructure consisting of a matrix of tempered martensite (α-Fe, BCC, strong spots, a0 = 2.866 Å), and particles of Fe3 C (cementite, orthorhombic, a = 4.524 Å, b = 5.088 Å, and c = 6.741 Å, weak spots). Index the pattern and establish the orientation relationship between the two phases. Assume a camera constant, λL = 46 mm Å. Notice that α-Fe appears to consist of two twin-related variants.

Solution: Based on Table 4.1.2, we can calculate the dhkl spacings for orthorhombic cementite as a table: hkl 001 010 100 011 101 110 002 111 012 102 020 112 102 dhkl (Å) 6.74 5.09 4.52 4.06 3.76 3.38 3.37 3.02 2.81 2.70 2.54 2.39 2.38 We can then index the diffraction pattern, using the camera constant, by measuring distances of the various reflections in the diffraction pattern, as follows:

011m

103t

10

112m

101

101 112 011

110

Trace of twin plane

lane Twin p m r no al

011 112 112t

112t

112m

OUP CORRECTED – FINAL

Interactions of Neutrons with Matter 535 The α-Fe matrix (indexed as m) is shown as a solid line and its twin (indexed as t) orientation as the dotted line. The cementite reflections are indexed without a subscript. This gives the orientation relationship:   (011)Fe3 C  011 α−Fe and 101 Fe

3C

 (112)α−Fe

Now to determine the zone axis, we take cross-products, (4.1.8), to give ¯ 1¯ ]Fe C  [31¯ 1¯ ]α−Fe (see Figure 4.1.10). [ 11 3

8.8 Interactions of Neutrons with Matter Neutron-based methods require access to national user facilities and are not routinely encountered in materials characterization and analysis; however, for the sake of completeness, we briefly discuss the scattering of neutrons by atoms. Even though neutrons are much more massive than electrons, carry no charge, and are electrically neutral, they possess a magnetic moment that can be polarized and contribute to scattering by magnetic materials. The interactions of neutrons with the nucleus is very short range and of the order of the nuclear radius, which is ∼10–4 times the atomic radius. As a result, most materials are not able to stop thermal neutrons and they penetrate deep (typically mms) into the specimens; a consequence of this is the requirement of a substantially larger quantity of a material for neutron scattering experiments.

Example 8.8.1: Calculate the wavelength of a thermal neutron (rest mass, m = 1.67 × 10–27 kg) emerging from a reactor at room temperature (300 K) with an energy of 1.5 kB T. Solution: We assume that the energy of the neutron is entirely kinetic. Thus 1 2 2 mv = 1.5 kB T. The wavelength, λ, is given by the De Broglie relation, λ = h/p = h/mv = h/(3m kB T)1/2 . Substituting the values of m, h, kB , and T, we get λ = 1.456 Å

8.8.1 Nuclear Interactions A neutron beam interacts with an atom in two distinct ways: first, with the nucleus using short-range nuclear forces, and second, with the magnetic moment of the atom via the spin of the neutron. 2 , Each atom of nuclear radius, RN , presents an effective surface area of 4π RN which is not penetrable by the incident neutron beam. Typically, RN is small

OUP CORRECTED – FINAL

536 Diffraction of Electrons and Neutrons compared to the wavelength, λn , of the thermal neutrons and is given by RN = 1.5 × A1/3 × 10−13 m, where A is the atomic mass number (see §5.3.5.1). 2 is called the scattering cross-section and gives the total The quantity 4π RN intensity of the neutrons scattered by the nucleus in all directions; however, because RN Bz

Post-Lens Field Br > B z

z vz Fr

Fθ Br

Rotation Out of Plane θ

F Bz vz

vθ Br Spiral Out Spiral In Nearly Helical Path

Figure 9.2.2 (a) The helical trajectory (exaggerated) of an electron in a magnetic field of induction, B, arising from the Lorentz force. (b) The electron trajectory through an electromagnetic lens causes both a rotation in the plane of travel by θ as well as a focusing of the electron along the optic axis at a position determined by the strength of the magnetic field. Adapted from Fultz and Howe (2013).

path as shown in Figure 9.2.2a. It is best to write the equation of motion of the electron subject to the magnetic field of the lens in cylindrical polar coordinates. The electron velocity, v = vr rˆ + vθ θˆ + vz zˆ , in the lens field of induction, B = Br rˆ + Bθ θˆ + Bz zˆ , will experience a Lorentz field with components F = Fr rˆ + Fθ θˆ + Fz zˆ = −evθ Bz rˆ + −e (vz Br − Bz vr ) θˆ + evθ Br zˆ

(9.2.1)

Note that for a lens with an axially symmetric field, Bθ = 0, is assumed. The electron trajectory is shown in Figure 9.2.2b. Initially, the electron travels in the plane of the paper with velocity, v, at an angle to the optic axis and experiences the prefield of the lens that is predominantly radial, i.e. Br >> Bz . As a result, it experiences a force, Fθ , pointing it out of the plane of the paper with a new velocity component, vθ , causing it to have a spiral motion. Now, when the electron with both vz and vθ components enters the lens field with Bz > Br , it now experiences a strong radial force, Fr , that focuses the electron beam. At the exact center (z = 0) of the symmetrical lens, vr = 0, vθ is a maximum, and the electron travels in a helical path with only vθ and vr components. As it traverses the exit half of the magnetic lens, because of a finite vθ component, the electron is now further focused; in addition, Br changes sign and vθ starts to decrease. By

OUP CORRECTED – FINAL

562 Transmission and Analytical Electron Microscopy symmetry, by the time the electron exits the post-field of the lens, vθ = 0, its spiral motion stops, and the electron comes to a focus along the optic axis. The field component, Bz , generated in the lens is directly proportional to the lens excitation current and hence the focal length decreases (vθ increases) as the lens current is increased. Note that unlike optical lenses the path of the electron beam is rotated out of the plane of the paper through an angle, θ , in radians given by 0.15 θ= √ E

 Bz dz

(9.2.2)

Axis

where E is the electron energy (eV). For 100 keV electrons passing through a field 0.15 4 of 1 T μ−1 0 , over a distance of 0.5 cm in the lens, we get θ = 3.16×102 ×10 ×0.5 = 2.4 rad. As a result, either the focusing of an image or a change in magnification, both of which involve changing the electromagnetic lens field, is accompanied by a rotation of the image about the optic axis. In addition, the angle subtended by the electron beam on the optic axis in an electron microscope is very small, i.e. of the order of 1◦ . Hence, compared to its angular spread, the path length of the electron beam is rather large and results in a very small effective numerical aperture, i.e. n sin α ∼ 10–2 . Earlier, in §6.8.2, we described the behavior of an optical lens in terms of geometric optics, with the assumption that the lens is sufficiently thin compared to the total optical path. We presented the simple relationship (6.8.3) between the focal length, and the position of the object and image. For electromagnetic lenses, this thin-lens approximation is strictly not valid; however, it is common practice to use geometric optics to illustrate the various imaging and diffraction modes of an electron microscope, and we shall also do the same in the sections that follow, knowing fully well that it is not strictly correct. The diffraction limited resolution, δD (nm), of an electromagnetic lens can be obtained to first order using the Rayleigh criterion, (6.8.2), with a small (α ∼ 0.57◦ ∼ 10 mrad) angular beam and the refractive index, n = 1, for vacuum, as δD =

0.61λ 0.61λ ≈ ≈ 60λ n sin α α

(9.2.3a)

where we have set sinα = α, for small α. We can write the wavelength, λ (nm) in terms of the electron energy, E (keV), corrected for relativistic effects as 0.03878 λ=    E 1 + 0.9785 × 10−3 E

(9.2.3b)

to give δD =

0.61 0.03878 0.61λ   =  α α E 1 + 0.9785 × 10−3 E

(9.2.3c)

OUP CORRECTED – FINAL

Elements and Operations of a Transmission Electron Microscope 563 This suggests a better resolution is to be expected at higher voltages. As a result, many high-voltage electron microscopes (HVEMs) with acceleration voltage up to 3MV have been constructed, but most of them have found limited use because of significant beam damage of the specimen (Table 5.3.2; Fig. 5.3.8). Instead, to achieve ultimate resolution, aberration corrected microscopes, §9.2.9, operating at 60–300 kV, are currently used. An example of the design and development of such an aberration-corrected TEM is discussed in [6]. Similar to an optical lens (Fig. 6.8.3), an electromagnetic lens also shows both spherical and chromatic aberration.Electron beams originating at the same point but incident on the lens at different distances from the optic axis are brought to focus at different points on the optic axis (spherical aberration). As a result, the image of a point on the optic axis appears as a disc of least confusion on the Gaussian image plane. For spherical aberration this disc of least confusion limits the aberrationlimited resolution, δS , which is related to the beam divergence, α, approximately as δS = 0.5CS α 3

(9.2.4)

where CS is the spherical aberration coefficient, and a critical parameter that defines the imaging performance of the electromagnetic lens. We can see that for any lens with a given CS , best performance is achieved when δD = δS , and observed at an optimal divergence angle, α opt , obtained by setting (9.2.3a) equal to (9.2.4) and given by  αopt =

0.61λ CS

1/4

 ∼ 0.9

λ CS

1/4 (9.2.5)

The angle, α opt , defined in practice by an aperture is determined by both the acceleration voltage (λ) and the lens characteristics (CS ). Example 9.2.1: Calculate the diffraction limited resolution and the optimum divergence angle for an electromagnetic lens with CS = 0.6 mm, in a TEM operated at 100 kV. Solution: From (9.2.5), we get the optimal divergence angle  αopt =

0.6 × 0.0037 × 10−9 0.6 × 10−3

1/4 = 8 × 10−3 rad.

From (9.2.3c), we get the diffraction limited resolution δD =

0.61 0.03877 = 0.28 nm. √   8 × 10−3 100 1 + 0.9788 × 10−3 × 100 1/2

OUP CORRECTED – FINAL

564 Transmission and Analytical Electron Microscopy The electromagnetic lens also suffers from chromatic aberration, which brings electrons originating from the same point but of different energies to focus at different points on the optic axis. This is because higher energy electrons are less deflected by the magnetic field of the electromagnetic lens (Fig. 9.2.2). Again, the chromatic aberration is also defined by a disc of least confusion at the Gaussian image plane that limits the resolution, δC , to δC = CC

E α E0

(9.2.6)

where E accounts for instabilities in both the source (gun) and lens currents, and the chromatic aberration constant, CC , is of the order of the focal length, f, for weak lenses, and CC ∼ 0.6f , for strong lenses. For typical electron microscopes with a thermionic source and E0 = 100 keV, E ∼ 1.5 eV, CC = f = 2mm, α ∼ 10 mrad, we get δC ∼ 0.3 nm. The spherical aberration of an electromagnetic lens can be corrected using pairs of magnetic hexapoles, and is discussed further in §9.2.9. In addition, a cone of electron rays from a point, P, on the specimen at some distance, y, from the optic axis is generally focused as an ellipse at the point, P , in the Gaussian image plane (GIP) (Fig. 9.2.3). This is because of the loss of axial symmetry in the lens resulting in a variation in the focal length for the two orthogonal axes, x and y, shown in the figure. This is called astigmatism (Fig. 9.3.20) and is an inherent feature in electron-optical lenses due to a residual asymmetry, or inhomogeneity, of the magnetization of the pole piece. Moreover, astigmatism is very sensitive to specimen effects including misalignment, asymmetry, and contamination. However, such twofold astigmatism is routinely corrected in a TEM by introducing additional sets of correction coils with variable magnetic fields, and placed orthogonal to each other about the optic axis. The current in the coils can be periodically adjusted during operation of the TEM to correct for astigmatism in the image (Fig. 9.3.20). Finally, the depth of field, df , of an electromagnetic lens, is the same as that defined earlier for an optical lens (Fig. 6.8.4). It is the axial distance over which the specimen can be moved and yet produce a focused image with resolution, δS , and is given by y Figure 9.2.3 A simple ray diagram illustrating astigmatism in a lens arising from a loss of axial asymmetry. As a result, the focal length varies about the optic axis with two principal lines of foci, along orthogonal directions, as shown.

GIP

P

x

P y-focus

x-focus

OUP CORRECTED – FINAL

Elements and Operations of a Transmission Electron Microscope 565 df = δS /α

(9.2.7)

The depth of focus, D, is the range of distances over which the image appears in focus in the image plane of the lens (Fig. 6.8.4). For a convergence semi-angle, α ∼ 10 mrad, and magnification, M, such that the effective aperture α  = α/M in the final image, the depth of focus is related to the depth of field: D=

δS M δS M 2 = M 2 df = α α

(9.2.8)

which is very large (see Example 9.2.2). Now that we have a general idea of the characteristics of an electromagnetic lens we can move on and describe the different sections of a transmission electron microscope, starting with the illumination section. Example 9.2.2: Is a typical TEM specimen with thickness less than 100 nm appropriate for imaging? Why? In addition, in a TEM the viewing screen and the image recording system are separated by 20 cm. For a magnification of 50,000×, can we both observe and record the image? Assume typical parameters for the TEM with a thermionic source. Solution: For a typical TEM with α ∼ 10 mrad, δ S ∼ 0.3 nm, we get df = 0.3 × 10−9 /10−2 = 0.3 × 10−7 m = 30 nm. The depth of focus is smaller than the specimen thickness and so it would be difficult to image it through its entire thickness. However, if α ∼ 1 mrad, df = 300 nm, and it should be possible to image it easily. The depth of focus,  2 (9.2.8), for α ∼ 10 mrad, gives D = M 2 df = 5 × 104 × 30 × 10−9 = 75 m! Clearly, the image will be in focus in both the viewing screen and the image recording system.

9.2.3 The Illumination Section The illumination section (Fig. 9.1.1) transfers the electron cross-over produced by the gun (Fig. 5.2.5) with diameter, d co , to the specimen. The cross-over serves as the object for the illumination system that consists of two or more CLs (called C1, C2, . . . ). Two illumination conditions—a broad parallel beam and a convergent beam focused on the specimen—commonly encountered in TEM are discussed next. Note that the beam is most coherent when the incident beam is parallel. 9.2.3.1 Parallel Beam Operation of a TEM To illuminate the specimen with a parallel beam of electrons, uniformly over a region of several microns, and magnifications ranging from 20,000× to 100,000×, a system of two CLs (Fig. 9.2.4) is used. For a thermionic source, with a large

OUP CORRECTED – FINAL

566 Transmission and Analytical Electron Microscopy (a)

(b) Source Cross-Over

Figure 9.2.4 Ray diagram illustrating two ways of producing “parallel” illumination on the specimen using (a) an under-focused C2 lens, and (b) a focused C2 lens to produce an image of the source cross-over at the front focal plane of the upper objective lens. The corresponding diffraction patterns produced in the back focal plane of the objective lens (Fig. 8.6.9 and Fig. 8.6.10) show sharp diffraction discs (very small diameter that depends on α) for (a), and very narrow spots for (b).

Condenser Lens, C1

C1 Cross-Over (demagnified source) C2 lens Under-Focused

α

Under-focused Beam

C2 Lens Focused Front Focal Plane of Objective Lens Upper Objective Lens

Specimen

cross-over diameter (Table 5.2.2) of several tens of micrometers, the first lens, C1, forms a demagnified image of the cross-over by an order of magnitude. On the other hand, for FEG sources where the cross-over diameter may be much smaller than desired for illumination, the lens C1 would be excited to magnify the crossover. Then, in either case, the second condenser lens, C2, is excited to produce a substantially under-focused image of the C1 cross-over on the specimen, with a convergence semi-angle, α < 10–4 rads (∼ 0.01◦ ; Fig. 9.2.4a), which for all practical purposes can be considered as a “parallel” beam. Such a parallel beam produces sharp diffraction patterns (very small discs) and the highest contrast in images. Note that typically, in this mode of imaging the C1 lens excitation remains unchanged (set at some value recommended by the manufacturer) and the C2 lens current determines the magnification (which is also related inversely to the area of the specimen illuminated by the beam). Alternatively, if a TEM is also used to generate a focused electron beam, e.g. in a STEM, additional control of the beam is made possible using the upper pole piece of the OL. In this arrangement, the C2 lens is focused to produce an image of the C1 cross-over at the front focal plane of the upper OL, which will then generate a true parallel beam on the specimen (Fig. 9.2.4b). 9.2.3.2 Focused Probe Formation for Illumination and Scanning A focused probe ( 0 (Fig. 9.3.6).

OUP CORRECTED – FINAL

596 Transmission and Analytical Electron Microscopy (a)

(b)

(c)

Incident Beam

Incident Beam (hkl)

(hkl)

2θ ghkl

θ

(hkl) 3θ Specimen

θ

Specimen 2θ

θ

ghkl

000

000

θ θ

Specimen θ

2θ

2ghkl

ghkl

ghkl

3ghkl

θ

000

Figure 9.3.5 Schematic illustration for setting up optimal two-beam diffraction conditions for DF imaging. (a) Standard condition. (b) Incident beam tilted through 2θhkl such that the diffracted beam, ghkl , is down the optic axis. In this case the diffracted beam is weak as indicated by the size of the diffraction spot. (c) By tilting the beam by −2θhkl , i.e. in the opposite direction, a strong higher intensity ghkl reflection is now along the optic axis, which makes for a good centered dark field image. 125 nm

g

Figure 9.3.6 Variation in diffraction contrast as a function of the excitation error: (a) s = 0, (b) s > 0, but small, and (c) s > 0, but large. The best contrast is observed in (b), even though the defect appears narrower in (c). Adapted (1996).

from

Williams

and

Carter

(a)

(b)

(c)

Diffraction contrast is widely used to resolve the microstructure of practical materials. Broadly speaking such microstructure features include changes in local orientation without changes in composition (grains, twins, precipitates, etc.), lattice defects (dislocations, stacking faults, etc.), and multiple-phase systems with additional changes in composition and/or structure, including their interfaces. Figure 9.3.7 shows diffraction contrast images of twins (see §8.7.6) formed in copper as a result of explosive deformation. A classic application of diffraction contrast is to determine the Burgers vector, b, of perfect dislocations, especially in isotropic materials. In particular, this

OUP CORRECTED – FINAL

Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods 597 (a)

(b)

(c) Figure 9.3.7 Twins in copper. (a) Bright field image, with (inset) the selected area diffraction pattern indexed. (b, c) Dark field images using the twin spots T1 and T2 , clearly identifying the twins by their contrast reversal.

T2 T1

T1 1μm

T2

Adapted from Thomas and Goringe (1979).

E

E A

E

A F

F D

A F

D

D

B

B C

C

C

0.2 μm

(a) g = (020)

B

(b) g = (200)

(c) g = (111)

analysis takes advantage of the feature of dislocation images that make them invisible in BF and DF images when the condition g·b = 0 is satisfied (§9.5.2). This is illustrated in Figure 9.3.8 for dislocations in aluminum. However, most materials are elastically anisotropic and the simple invisibility criterion cannot be applied; more sophisticated analysis is required and the interested reader may consult Edington (1975), Thomas and Goringe (1979), or Williams and Carter (1996) for details. Diffraction contrast is observed in both TEM and STEM images, even though in the latter case, the contrast is a lot poorer. To get good diffraction contrast images in TEM, we need a parallel incident illumination, a strong two-beam diffraction condition, and allow only one of the two beams, either transmitted (BF) or diffracted (DF) beam, to pass through the OA and form the image. In STEM, as we have seen, we require a large collection angle to generate sufficient signal intensity in the detector. However, by the reciprocity principle, we require this collection angle in STEM to be as narrow as possible to approximate the parallel illumination condition in the TEM. Hence, diffraction contrast in STEM is only possible for small collection angles, which naturally reduces the image intensity and makes it noisy (Fig. 9.3.9). Some of this limitation can be overcome using a high brightness FEG source, but diffraction contrast is not the forte of STEM.

Figure 9.3.8 Images of the dislocations in aluminum taken with different operative reflections. Note that the dislocations marked D and E (encircled) can be observed only for g = (020). Applying the g · b = 0 criterion gives the Burgers vector b = 12 [011] for these dislocations. Adapted from Edington (1975).

OUP CORRECTED – FINAL

598 Transmission and Analytical Electron Microscopy

(a)

(b)

(c)

0.5μm Figure 9.3.9 Comparison of diffraction contrast in STEM and TEM images. The same region of an Al-4 wt% Cu specimen is imaged in bright field STEM under (a) normal and (b) narrower collection angles; in the latter case, some diffraction contrast (bend contours) is visible, and (c) the corresponding TEM BF image. In all images, the Cu-rich precipitates show excellent mass-thickness contrast. Adapted from Williams and Carter (1996).

9.3.4 High-Angle Incoherent Scattering: Z-Contrast Imaging The Coulomb interaction of the fast electron beam with the positive potential (+Ze) of the atomic nucleus is strong enough to scatter the electrons into high angles (>3◦ ). In fact, the electrons can even be back-scattered, and then detected for imaging in a scanning electron microscope (SEM; §10.3.3). The Rutherford cross-section, (9.3.11), applicable for such scattering, depends strongly on the atomic number (∼Z 2 ), and offers the possibility of chemical contrast in these images. In other words, areas and particles of high-Z elements scatter strongly and produce bright contrast in images recorded with electrons scattered into high angles (Fig. 9.3.10); in particular, this effect is used in HAADF imaging in a STEM as schematically illustrated in Figure 9.3.11. Note that if a small detector is used, the collection angle of the HAADF detector has to be large enough to avoid detection of Bragg scattered or diffracted electrons. However, if we use a detector large enough to collect many Bragg reflections, then we get an incoherent image [110, 111]. This ensures that the detector collects purely incoherent but elastically scattered electrons that are described well by the Rutherford scattering model, (9.3.11). Optimally, such a detector would collect electrons scattered—incoherently—through angles larger than 70 mrad (∼8◦ ). As a result, the intensities of scattering from individual atoms can be added together and the images can be interpreted directly in terms of the

OUP CORRECTED – FINAL

Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods 599

(a)

500 nm

(b)

500 nm

Figure 9.3.10 Images of iron oxide nanoparticles drop cast on ultra-thin carbon films mounted on Cu grids at nearly identical magnifications. (a) A bright field TEM image showing dark contrast for the particles with strong diffraction contrast—notice how some particles are darker because they are oriented crystallographically for strong Bragg scattering. (b) A dark field HAADF–STEM image, with contribution from only incoherent scattering over sufficiently large angles (>70 mrads) such that the particles provide bright contrast. Compare with Figure 9.2.18, where the DF image in the STEM, collected with an ADF detector, includes strong contributions from diffracted electrons, which by definition has significant coherent scattering contributions. Images courtesy Dr. R. Hufschmid.

Electron beam focused to atomic dimensions Specimen (GaAs) Z=31 Z=32

Ec

Intensity

HAADF

Ga As 1.4Å

Spectrometer

Ev 620

660

EELS (eV)

Figure 9.3.11 Schematic illustration of a STEM instrument with a HAADF detector for atomic resolution, Z-contrast imaging, and an EELS spectrometer for simultaneous chemical and electronic structure measurements of the specimen. GaAs image courtesy of Professor S. J. Pennycook.

OUP CORRECTED – FINAL

600 Transmission and Analytical Electron Microscopy atomic species (Z) and their position. Hence, the use of the name, Z-contrast imaging, to describe this technique. Now, we briefly discuss the achievable resolution in such Z-contrast imaging. Typically, the spatial extent of an atomic scattering potential is ∼0.01–0.03 nm, and modern dedicated STEM instruments produce probe sizes of the order of 0.07–0.20 nm, with the lower limits achieved in aberration corrected instruments operating at 200 kV and generating about 30 pA of current. As a first-order approximation, a vertical column of atoms encountered by the probe in a thin foil specimen can be considered as a δ-function potential. An ultimate STEM, using Z-contrast imaging with HAADF detectors, can produce an image resolution that is a convolution of the atomic potential with the spatial distribution of the electron current in the probe. Hence, for aberration-corrected STEM instruments with sub-Å probe diameters, individual atomic columns can be resolved (Fig. 9.3.12). The success of such aberration corrected STEM instruments in producing atomic resolution images, has generated interest in using lower acceleration voltages to avoid knock-on damage, or atomic displacements, of specimens (§5.3.2.2) during imaging. However, at lower voltage, chromatic aberration becomes the limiting factor for spatial resolution in Z-contrast imaging in a STEM [20]. To overcome this limitation, various monochromator designs have been incorporated in an aberration corrected STEM [20–23]. It is logical to see that the use of a HAADF detector also allows the placement of an energy-loss spectrometer along the optic axis of the microscope, in place

(a)

O Ti (b)

Figure 9.3.12 (a) HAADF Zcontrast image of a Co-doped TiO2 film, grown epitaxially on LaAlO3 substrate. The raw data, shown in the inset, has been smoothed to remove noise. Notice the bright contrast from the heavier La in the substrate on the left. (b) Shows a schematic of the TiO2 anatase lattice, with grey and red spheres showing Ti and O atoms, respectively, along the projection of the image. Adapted from [24].

[001]

1 nm

OUP CORRECTED – FINAL

Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods 601

LaFeO3 Spectrum Image EELS

OK

Ti L

Fe L

La M EDXS

2 nm

SrTiO3 OK

Ti K

Fe K

La L

Sr L

of a BF detector, to collect the inelastically scattered EELS (Fig. 9.3.11). By synchronizing the EELS acquisition with the raster scan of the probe, each pixel in the high-resolution STEM image can be associated with a specific EELS signal, providing simultaneous structural, chemical, and bonding information of the specimen, all at atomic resolution. EELS is discussed further in §9.4.2, but Figure 9.3.13 illustrates the power of this technique. In summary, a HAADF detector in a STEM uses an Å-size probe and collects incoherently scattered electrons at large angles (∼70–150 mrads) to produce high-quality Z-contrast images at atomic resolution. By the reciprocity theorem, to accomplish the same in a TEM will require equivalent large angles of beam convergence, which is rather difficult, if not impossible. Even when the so-called hollow cone illumination [109] is used in a TEM, we can only have convergence semi-angles of a few mrads, making it impossible to obtain pure Z-contrast images. It is inevitable at the small convergence semi-angles prevalent in the TEM that some diffraction contrast is included in the image.

9.3.5 High-Resolution Electron Microscopy (HREM): Phase Contrast Imaging in Practice Phase contrast images are best formed when a thin crystal is oriented such that the incident electron beam is parallel to a low-index zone axis and a fixed number of diffracted beams are selected by the objective aperture to form the image

Figure 9.3.13 Simultaneously acquired atomic resolution, Z-contrast images, EELS, and the much noisier energy dispersive X-ray spectra (EDXS) maps from a LaFeO3 /SrTiO3 interface. Adapted from [25], which is also an excellent review of STEM with a historical perspective.

OUP CORRECTED – FINAL

602 Transmission and Analytical Electron Microscopy (Fig. 9.2.10d). Generally, the illumination is called axial if the transmitted or forward-scattered beam is along the optic axis, but if it makes an angle with the optic axis, it is called off-axis illumination. Such phase contrast images produce a pattern of bright and dark lines or dots arising from the interference of the diffracted beams with the direct beam and reflecting the periodicity of the crystal lattice orthogonal to the projection direction. In the simplest form of phase contrast imaging, the interference of any two beams passing through the back focal plane of the OL and selected by the objective aperture gives rise to lattice fringes in the image. Such lattice fringes are formed even when the incident beam is not exactly aligned along a specific crystallographic direction of the specimen, and are observed over a wide range of lens defocus and specimen thickness values. In fact, in a specimen containing nanoparticles, or grains, with a Debye–Scherrer ring diffraction pattern, it will be impossible to align the beam parallel to a specific crystallographic direction in all nanocrystals. Nevertheless, even in specimens with a distribution of nanocrystal orientations, lattice fringes and even individual atomic columns can be observed (Fig. 9.3.14) without any careful alignment of the diffraction conditions, provided

220

(a)

(e)

(i)

100 nm

100 nm

(c)

(b)

1 μm

(f)

(m)

311 400

100 nm

(j)

10 nm

(g)

(k)

100 nm

100 nm

10 nm

(d)

(h)

(l)

100 nm

100 nm

10 nm 10 nm

Figure 9.3.14 (a) A low-magnification image of a high purity and monodisperse iron oxide nanoparticles dispersed on a carbon grid. (b) A selected area diffraction pattern from the same region showing a ring pattern corresponding to the magnetite phase. (c)–(h) Images of the same region of the nanoparticles taken at increasingly higher magnifications. (i)–(l) phase contrast images, centered on the same particle in strong diffraction contrast, at increasing magnifications. (m) A high-resolution phase contrast image of a particle, aligned to the [310] zone axis, taken on a CS -corrected 300 kV TEM, at a defocus of 213 nm showing lattice fringes of individual atomic columns. Courtesy Dr. R. Hufschmid [26].

OUP CORRECTED – FINAL

Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods 603 (a)

4.15 nm

(c)

2.03 nm

32H

(b)

2.4 nm

c = 82.82 Å

(d) 27R

16H

c = 72.0 Å

c = 40.7 Å

0,N Al

Figure 9.3.15 Polytypes in the Al-O-N system are formed by the insertion of a chemically distinct layer into another chemically distinct structure. Onedimensional lattice images observed in polytype structures of (a) 32H where the unit cell consists of two blocks, each containing sixteen layers, (b) 27R with three blocks of nine-layer repeats, and (c) 16H with two blocks of eight-layer repeats. (d) A schematic of the projections of the unit cell on the (110) planes of the three polytype structures. The 001-diffraction row for each structure is shown in the inset in (a)–(c). Adapted from [27].

the reflections excited correspond to lattice spacing larger than the resolution of the microscope. Generally, since the diffraction conditions are not exactly specified, lattice fringes provide information only about the morphology, shape, and select inter-planar spacing in such nanocrystals. Next, we have what are known as one-dimensional structure images, typically observed in long-period unit cell crystals. By tilting the orientation of the specimen, the incident beam can be made orthogonal to the long-axis of the crystal unit cell. In this case a systematic row of diffracted spots in one dimension, symmetric about the transmitted beam, can be excited. Phase contrast images with this systematic row of diffraction spots can provide additional information, albeit in one dimension, about the crystalline specimen. Figure 9.3.15 is such a one-dimensional structure image of polytype6 structures (see Verma and Krishna, 1966) formed in the ternary Al-O-N system. Strictly speaking, an HREM image contains no more structural information than the underlying diffraction pattern. If the incident beam is parallel to a specific crystallographic axis, typically of low index, and satisfies the Bragg condition in two dimensions, the diffraction pattern will reflect the symmetry and lattice parameters of the crystal system and unit cell, respectively. In this case, a phase contrast image formed by the interference of the transmitted and diffracted beams

6 Forms of a crystalline substance that only differ in one dimension of the unit cell.

OUP CORRECTED – FINAL

604 Transmission and Analytical Electron Microscopy will produce a two-dimensional lattice fringe image, representative of the unit cell but not specifically the location of the individual atomic columns within it. Figure 9.3.16 shows such a two-dimensional image of β-Si3 N4 , with the incident beam along [110], and including (002) and (111) reflections, in addition to the transmitted beam. This image is rich in detail and includes many types of defects including dislocations, stacking faults, as well as twin and tilt grain boundaries. From a practical point of view, since such two-dimensional lattice images are formed with a small number of diffracted beams, the images can be formed over a

a

s

b d

c

e m k

l

j

s s

[001] i [001] Figure 9.3.16 Two-dimensional lattice image of β-Si3 N4 with incident beam parallel to [110]. Image shows a number of defects including dislocations (b–c, d–e), stacking faults (s), tilt grain boundaries (f through m), and twin boundaries indicated by arrowheads. Adapted from Shind¯o and Hiraga (1998).

[110]

h

[110]

g s

f 3 nm s

OUP CORRECTED – FINAL

wide range of defocus (including the Scherzer value) and thickness. Again, from these images it is not easy to say whether the atomic columns appear as black or white dots in these images. However, to analyze defects and their structural details, these types of images should be taken as close to the Scherzer defocus condition as possible; moreover, very thin specimens (∼5–10 nm) will facilitate direct interpretation of the images of defects. The most sophisticated images in HREM are those that represent atomic positions as bright or dark regions in a two-dimensional structure image. However, true structure images are obtained only in very thin specimens (∼8–10 nm thick), where the amplitude of the various diffracted beams are proportional to the thickness. Figure 9.3.17 shows the amplitude of various diffracted beams in β-Si3 N4 ; up to 7 nm in thickness they all show a linear dependence with thickness. The corresponding HREM images (Fig. 9.3.18) simulated for various defocus values show best structure images are found around the Scherzer defocus (48.7 nm). Further, the optimal thickness to be used depends on the size of the unit cell, even if they are of similar densities. Figure 9.3.19 illustrates that the structure image of α-Si3 N4 can be best obtained for specimens twice the thickness of β-Si3 N4 . In practice, recording of optimal HREM images require that the TEM is corrected for astigmatism and specimen drift is avoided during the time of image acquisition. Both can be accomplished by recording a high magnification image of the amorphous carbon layer, typically found at the edge of the specimen. A Fourier transform or diffractogram of the image will give a good indication of the misalignment, astigmatism, and specimen drift (Fig. 9.3.20). Astigmatism typically causes the diffractogram to look elliptical instead of circular, whereas drift causes the disappearance of parts of the rings along the drift direction. Such astigmatism can be corrected in a TEM using the pair of stigmators—a set of two magnetic quadrupole lenses, one above the other, and rotated by 45◦ —positioned above the condenser lens, C1, to produce a circular incident beam, and above the OL to produce a focused, minimum-contrast image.

(a)

(g)

(b)

(h)

(c)

(i)

(d)

(j)

(e)

(k)

Amplitude

Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods 605

0.2

0.1

0

10

20 30 40 Thickness (nm)

50

60

Figure 9.3.17 Dependence of the amplitudes of different diffracted beams as a function of specimen thickness calculated for a 400 kV TEM. Adapted from Shind¯o and Hiraga (1998).

(f)

(l)

Figure 9.3.18 Simulated highresolution images of a thin (3 nm) specimen of β-Si3 N4, for [001] incidence and 400 kV, as a function of defocus varying from −40 nm to +70 nm in steps of 10 nm. The structure images (Fig. 9.3.19) appear close to Scherzer defocus (f = 30 − 50 nm). Compare with Figure 1.4.14. Adapted from Shind¯o and Hiraga (1998).

OUP CORRECTED – FINAL

606 Transmission and Analytical Electron Microscopy (c)

(a)

(d)

1 nm (b)

Figure 9.3.19 Structure images of (a) β-Si3 N4 , and (b) α-Si3 N4 . The corresponding simulated images, assuming a thickness of 3 nm and defocus, f = 45 nm, with and without the atom positions superimposed are in (c, d) and (e, f), respectively. Adapted from Shind¯o and Hiraga (1998).

7 Fresnel fringes can also be used for focusing. Typically, a white fringe means under-focus, and so raising the specimen or increasing objective lens strength will get the image into focus.

(e)

(f) 1 nm

Once the astigmatism and specimen drift are eliminated, a good HREM image can be recorded, provided steps are taken to ensure that the microscope is well aligned, an optimal thin area of the specimen has been identified, and the optics are set up with the required diffraction pattern. Finally, it is required to set the optimum defocus. To do so, again an amorphous region of the specimen is used to first find the focus condition that produces a minimum contrast condition, i.e. when nearly all details in the image of the amorphous material disappear.7 In this case, the CTF is close to zero over a wide range of spatial frequencies and occurs at a defocus, fmc = −0.44(CS λ)1/2 . For a given microscope, with a given CS , the minimum contrast occurs at a specific value of defocus, e.g. for a JEOL 4000EX microscope, it is at fmc = −18 nm. From here, a specific number of defocus steps would take one to the Scherzer defocus, which for the same microscope is at fSch = −49 nm. To avoid any errors in defocus setting, it is common practice to record images over a range of focus values (through focus series). HREM structure images have two significant limitations: they provide very little information on the atomic number of the elements in the specimen, and they reveal only a projection of the atomic columns in the crystal structure

OUP CORRECTED – FINAL

Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods 607 (d)

(a)

(b)

(e)

(c)

(f)

Figure 9.3.20 Images of an amorphous carbon film and their diffractograms under different imaging conditions using a 300 kV HREM. (a) Well-aligned, no drift, astigmatism corrected. (b) Some astigmatism. (c) Significant astigmatism. (d) No astigmatism but with small drift (0.3 nm). (e) No astigmatism but with larger drift. (f) Well-aligned, no drift, and no astigmatism with graphite fringes of 0.344 nm spacing. Adapted (1996).

0.5 Å–1 along the incident beam direction. Moreover, they require very thin specimens for direct structural interpretations, and in many cases where there may be significant surface reconstruction (§8.3.1), thicker specimens may be necessary to reveal the true structural details. In the latter case, inelastic scattering of electrons in the specimen will deplete the elastic wave field, which in turn, will affect the phase contrast images for specimens of thickness greater than a few tens of nanometers. In §9.4 we briefly introduce inelastic scattering and most importantly, present two important spectroscopy methods that arise from inelastic scattering that complement imaging and, with the incorporation of appropriate spectrometers, provide quantitative chemical and electronic structure information of the specimen in an (analytical) electron microscope. Example 9.3.1: A thin film of NiO (a = 4.18 Å) includes small precipitates of Ni (a = 3.52 Å) that are perfectly aligned, i.e. (001)NiO || (001)Ni and [100]NiO || [100]Ni . In a TEM experiment a diffraction pattern is taken such the beam direction is along the [001] direction of the crystals. Then a twobeam lattice image is also produced. (a)

Draw the (001) zone axis pattern for NiO and Ni.

(b)

What would the diffraction pattern look like if double diffraction is included?

(c)

Describe the two-beam lattice image including double diffraction.

from

Williams

and

Carter

OUP CORRECTED – FINAL

608 Transmission and Analytical Electron Microscopy Solution: We refer to the following figure to answer the questions. (a)

(b)

020

(c)

220 gNiO gNi

000

(d)

200

(e)

(f)

(g)

dtm

gNiO gNi gtm

(a)

The (001) zone axis pattern for the two crystals, with the reflections indexed, are drawn to scale. NiO has the larger lattice parameter and since it is the matrix has brighter intensity (diameter of the circles).

(b)

In double diffraction, each diffracted spot of the precipitate (Ni) acts as an incident beam for the matrix phase (NiO). In other words, the two diffraction patterns are convoluted. A simple way to visualize this is to separate several matrix patterns (two shown) displaced by diffraction vectors of the precipitate. This generates all the double diffraction patterns.

(c)

An experimental diffraction pattern, adapted from Williams and Carter (1996), from perfectly aligned Ni and NiO, agrees very well with (b).

(d)

Lattice fringes are produced by the interference of transmitted and diffracted beams. Similarly, we can also produce an interference between two beams that are diffracted from the matrix and precipitate. Such fringe patterns are called Moiré patterns and the simplest kind, called a translational Moiré, involves lattice planes that are parallel, but with different spacings. The two reciprocal

OUP CORRECTED – FINAL

Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods 609 lattice vectors, gNiO and gNi , when perfectly aligned will effectively generate another spacing defined by gtm = gNi – gNiO , with spacing given by 1/gtm in the lattice image. (e)

The lattice image of NiO, and (f) the lattice image of Ni, showing the (200) lattice planes to scale.

(g)

The translational Moiré pattern produces an additional period, d tm , where dtm = g1tm = dNiO = 2.09 1.76 = 13.24 Å, as illustrated in dNi 1− d

NiO

1− 2.09

the figure. Note that Moiré patterns can also be produced by the rotation of one crystal with respect to the other. .

9.3.6 Magnetic Contrast: Lorentz Microscopy The two conventional modes of Lorentz8 microscopy, Fresnel9 and Foucault10 imaging, are now well established. In Fresnel imaging, a small defocus, z, is introduced, and contrast arises wherever there is a spatially varying in-plane component of magnetic induction. Fresnel contrast can be understood based on a classical ray diagram (Fig. 9.3.21a). The electrons on passing through the (a)

8 Hendrik Antoon Lorentz (1853– 1928) was a Dutch physicist who shared the Nobel prize (1902) in Physics with P. Zeeman and was cited for “their research into the influence of magnetism upon radiation phenomenon.” 9 Augustin-Jean Fresnel (1788–1827), French physicist who contributed to early developments in optics. 10 Jean-Bernard-Leon Foucault (1819– 1868) was a French physicist.

(b)

t B⊥ Δz

Intensity

Convergent Divergent 10 μm Position, x Figure 9.3.21 (a) Schematic illustration of the contrast in a Fresnel image, obtained with a defocus, z, of two ferromagnetic domains separated by a 180◦ domain wall. Note that even if there is a net deflection by the magnetic specimen, in a focused bright field image it will not show any magnetic contrast. (b) Example of a Fresnel image of domains in a 30-nm thick cobalt film. The magnetization ripple contrast allows for interpretation of the direction of magnetization in the interior of the domains [29].

OUP CORRECTED – FINAL

610 Transmission and Analytical Electron Microscopy

11 Magnetization ripple is a characteristic fluctuation of magnetization always oriented perpendicular to the average magnetization direction of the domain.

specimen are deflected by the Lorentz force arising from its in-plane induction integrated over its thickness. In the example, the two domains with antiparallel directions of induction will deflect the beam in opposite directions. At any plane not coincident with the specimen plane, defined by the defocus, z, there will be regions beneath the domains where the electron intensity will be enhanced or reduced above a uniform background signal. Hence, domain walls are revealed as narrow dark or bright bands, on a uniform background with no contrast from the interior of the domains (Fig. 9.3.21b). Based on this simple description of Lorentz deflection, Fresnel images of domain walls arising from both the convergent (bright) and divergent (dark) electrons should be uniform and devoid of any additional fine structure. However, the narrow bright band corresponding to the convergent wall reveals additional fringes that arise from the interference of the waves from the neighboring regions of uniform but opposing directions of magnetic induction. Details of the intensity distribution of the fringes, if carefully analyzed, using models of the magnetization distribution in the walls, combined if necessary, with micromagnetic simulations, can provide additional information on any induction variation within the wall itself. Fresnel imaging is characterized by operational simplicity, high contrast, and no directional preference for imaging domain walls. However, since there is no contrast from the interior of the domains, in the absence of magnetization ripple11 it is difficult to determine the local direction of induction. In contrast to Fresnel imaging, the Foucault mode is an in-focus method of domain imaging and relies on a deliberate shifting of the objective aperture to generate magnetic contrast (Fig. 9.3.22a). If the objective aperture is positioned in the back focal plane such that scattering vector components, kx < 0, are obstructed and kx > 0 are allowed to transmit unchanged, an image as shown will be recorded. In this simple illustration, using a classical description, only those electrons with positive value of the deflection angle, β x > 0, contribute to a bright image. Thus, domains that deflect electrons in a well-defined direction, selected by the position of the objective aperture, will appear as high-intensity, bright areas in the image. All other electrons scattered outside the objective aperture will appear as zero intensity. Thus, manipulating the position of the objective aperture allows the imaging of domains and the qualitative determination of the induction along any given direction (Fig. 9.3.22b). Fresnel and Foucault methods are complementary; the former determines the location of domain boundaries and the latter provides information on the directions of in-plane induction inside specific domains. However, under standard operation both methods do not provide a quantitative description of the spatial variation of the induction in the specimen. The alternative method of differential phase contrast (DPC) imaging is best implemented in a STEM equipped with a large circular electron detector that is split into four quadrants (Fig. 9.3.18a). In the STEM, a small probe of electrons, often from a coherent source, is scanned across the specimen and the scattered electrons are detected to form the image in a time-sequential manner. The

OUP CORRECTED – FINAL

Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods 611 (a)

(b) B⊥ t Objective Lens

Objective Aperture

Image Plane Figure 9.3.22 (a) Schematic illustration of the image formation in a TEM in the Foucault mode. Notice that the interior of the domains is imaged. (b) Foucault images of a micron-scale magnetic element along two perpendicular scattering directions allow the unambiguous determination of the magnetization direction in the domains. Images courtesy of Professor J. N. Chapman.

detector, of fixed size, is positioned in the far-field with respect to the specimen. A set of post-specimen lenses between the specimen and detector ensure both a variable cameral length and the collection of electrons over a wide angular range of scattering. In this arrangement, the DPC image is the difference in signal between the two halves or opposing quadrants of the detector [30, 31]. Figure 9.3.23a shows a classical representation of the formation of DPC images and its relationship to the magnetic structure. The focused probe is scanned across the specimen, and at each point it is deflected by the local Lorentz angle: eλ βx (x) = h

+∞ 

By (x, z) dx = By (x) −∞

eλt h

(9.3.13)

where By (x) is the average in-plane induction along the trajectory of the electrons. It then passes through the descan coils before arriving at the detector to ensure that, in the absence of a specimen, the beam is centered and the number of electrons arriving at each sector of the detector is the same and independent of the position of the probe on the specimen. Thus, if the probe undergoes a deflection, β, at any point on the specimen as shown, the electrons falling on the opposing sectors of the detector would be different. The difference in signal would be proportional to β in both magnitude and direction. Further, by scanning the probe over the specimen surface, the deflection at every position, β(x,y), can

OUP CORRECTED – FINAL

612 Transmission and Analytical Electron Microscopy (a)

Probe-Forming Aperture Scan Coils

(b)

α Specimen β Descan Coils Post-Specimen Lenses 200 nm Quadrant Detector Figure 9.3.23 (a) Schematic arrangement of the implementation of the differential phase contrast imaging in a STEM. The intensities measured in each of the four quadrants of the detector provides a quantitative map of the magnitude and direction of the beam deflection at each position, proportional to the in-plane induction, as the probe is scanned across the specimen surface. (b) Representative DPC images of a cobalt magnetic element. Two sets of difference images, a vector map of the magnetization and the sum of the intensities in the four quadrants, equivalent to a conventional TEM image are shown. Images courtesy of Professor J. N. Chapman.

be mapped. A quadrant detector is used to resolve both components of the inplane induction (Fig. 9.3.23a). Since this is an in-focus method, if the signal from the four quadrants is added together, instead of taking the difference, a normal BF image of the specimen, revealing any non-magnetic features (such as defects, grain boundaries, etc.) of interest, is also produced. If a STEM with a quadrant detector is not available, a DPC image can also be produced in a conventional TEM. Using reciprocity arguments, we see [32] that DPC contrast maps of the in-plane induction can be obtained in a conventional TEM by digitally combining a series of Foucault images taken with small increments of the beam tilt in two orthogonal directions. From a practical point of view, DPC imaging in a STEM can provide resolutions of the order of ∼10 nm. It is very sensitive and is able to detect small quantities such as a few layers of Fe. However, all TEM imaging requires electron transparent specimens; often, this entails thinning of the specimen and

OUP CORRECTED – FINAL

Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods 613 this procedure may affect the domain structures to be observed. Also, typically electron-optical lenses generate strong magnetic fields that may affect/alter the magnetic structure of the specimen during imaging. There are three ways of avoiding the effect of the fields generated by the excitation of imaging lenses— turn the OL off or move the specimen away from the lens. Both these methods degrade the resolution of the instrument. Alternatively, dedicated field-free lenses specially designed for magnetic imaging can be installed. Finally, in DPC imaging due to the contribution from microstructural features the scattered pattern may not be symmetric and may sometime lead to pronounced contribution to the phase contrast image. To suppress the effect of such scattering, a ring detector (essentially an octant detector), instead of a quadrant, is used to exclude the contribution from the central beam [33].

9.3.7 Electron Holography Even though electron holography was originally proposed [33,34] to correct for microscope aberrations as a means to achieve superior resolution, its usefulness for magnetization measurements [34], recognized immediately afterwards, is emphasized here. The applications of electron holography to magnetic materials are based on the recording of an interference pattern from which the amplitude and phase of the object is reconstructed. Magnetic materials are strong phase objects and the phase shift of electrons passing through a magnetic specimen is proportional to the flux enclosed. For example, two beams starting from a coherent point source, passing through a thin ferromagnetic specimen (Fig. 9.3.24a) at two different points, P1 and P2 , and then brought to the same observation point on the image plane, would undergo a phase difference, φ, due to the enclosed flux given by

φ(x) =

et 

x2 By (x)dx

(9.3.14)

x1

where x1 , x2 are the positions of P1 , P2 , respectively, and By (x) is the average value of the in-plane induction of the specimen along the electron trajectory. Naturally, if P1 and P2 were to lie on the same magnetic line of force in the film, the phase difference would be zero. Therefore, magnetic lines of force may be directly observed as contour maps of the electron wave propagation by electron holography. This simple interpretation is valid provided the phase shift is entirely magnetic in origin. In practice, an electron biprism (typically, a charged wire placed between two earthed plates) is required to split the incoming electron wave into two virtual coherent sources and form the interference pattern. Even though the phase difference between any two arbitrary points in the specimen plane can be defined in this way, in the most commonly implemented form of electron

OUP CORRECTED – FINAL

614 Transmission and Analytical Electron Microscopy

(a)

Coherent Electron Source

(b)

Coherent Electron Source (FEG)

Imaging Lens

Condenser Lens Magnetic Specimen Objective Lens

Magnetic Specimen (Film)

P1

Objective Aperture

P2 Electron Biprism

Electron Biprism

Phase Difference, φ

Imaging Plane

Imaging Plane Hologram

Figure 9.3.24 (a) Principle of electron holography. Two beams from a coherent source converging at the same observation point after passing through a magnetic specimen, with in-plane induction as shown by arrow, would undergo a phase difference proportional to the enclosed flux. (b) Schematic drawing of the experimental arrangement for off-axis holography imaging in a TEM. A charged wire serves as the biprism splitting the beam into a reference wave (light) and a wave that passes only through the specimen (dark) and causes a phase change. The interference of the two waves produces the hologram. Adapted from Krishnan (2016).

holography, a vacuum or a reference wave that avoids the specimen completely is made to overlap with the wave scattered by the specimen to form the hologram (Fig. 9.3.24b). This configuration is termed off-axis or side-band holography and is one of at least twenty different ways identified to implement electron holography [35] in a TEM. Finally, the magnetic phase image is reconstructed from the hologram by on-line digital processing methods [37] to achieve truly quantitative electron holography. The resolution of the reconstructed image depends on the spacing of the fringes in the hologram. For magnetic imaging, a large number of fringes are desirable, and as large as 500 fringes have been reported [36]. To accomplish this, electron sources with high spatial coherence are desirable. High brightness, spatially coherent, field emission gun sources [38], now routinely available on commercial TEMs, greatly facilitate the implementation of electron holography. The principles of off-axis electron holography are discussed next; however, the monograph by Tonomura (1999) and a recent review [116] are comprehensive and contain more details of all aspects of electron holography.

OUP CORRECTED – FINAL

Beam–Solid Interactions, Contrast Mechanisms, and Imaging Methods 615 Figure 9.3.24b shows the experimental set-up for holography in a conventional TEM with a coherent FEG source. An electrostatic biprism is used to overlap the object (interacting with the specimen placed such that it covers about half the field of view) and reference wave (the other non-interacting half) to form the interference pattern. Ideally, the biprism would be rotatable, in plane, to facilitate the alignment of the interference fringes with the microstructural features of interest in the specimen. Quantitative analysis of phase shifts is carried out by acquiring data with a linear detector having a large dynamic range such as a CCD camera. The intensity distribution in the hologram, using coherent reference and object wave functions,  r and  0 , respectively, is given by  2   I (x, y) = |r (x, y)|2 + 0 x, y + | r (x, y) 0 (x, y) | ei(φ1 −φ2 ) + e−i(φ1 −φ2 ) = A2r + A20 + 2Ar A0 cos φ

(9.3.15)

where φ is the phase angle and A is the amplitude of the waves. When this hologram is recorded, a series of cosine fringes are superimposed on a normal TEM BF image (Fig. 9.3.25a). A Fourier transform of the hologram (Fig. 9.3.25b) reveals two complex conjugate side bands that, in addition to the fundamental cosine frequency, contain additional intensity distributions that include complete information on the amplitude and phase of the waves. The separation of the side bands can be controlled by the relative angular deviation of the object and reference waves or, in practice, the voltage applied to the biprism. Only one of the side bands is necessary for the reconstruction process. To reconstruct the hologram, one of the side bands is extracted, re-centered. and then its inverse Fourier transform is calculated. The resulting image is complex and using the real (r) and imaginary (i) data sets, the phase (φ) and amplitude (A) can be calculated as   i φ = tan−1 r  A = r2 + i2

(9.3.16)

This is shown in Figure 9.3.25c,d. In one dimension, the change in phase of the object wave incident in the zˆ direction, on a specimen in the x-y plane, is given by  φ(x) = CE

V (x, z) dz −

e 

 B⊥ (x, z)dxdz

(9.3.17)

where, V(x,z) is the mean inner potential and B⊥ is the component of the magnetic induction perpendicular to both xˆ and zˆ . Here, CE is given by CE =

2π E + E0 λE E + 2E0

(9.3.18)

OUP CORRECTED – FINAL

616 Transmission and Analytical Electron Microscopy (a)

(b) sideband

(e)

(f)

100 nm 7

100 nm

(c)

(d) 6 5 4 3 2

200 nm

1

Figure 9.3.25 (a) Off-axis hologram for a chain of magnetosomes. (b) Fourier transform of the hologram showing the side bands, one of which is used for the reconstruction. (c) The phase image, and (d) the amplitude image. (e) Bright field image of a double chain of magnetite magnetosomes, with white arrows representing [111] crystallographic directions. (f ) Magnetic phase contours measured using off-axis holography from two sets of magnetosomes, after magnetizing parallel and antiparallel to the arrow. The color wheel shows the direction of magnetic induction. Contours of spacing 0.25 radians can be seen. (a–d) Adapted from [39]. (e–f ) Adapted from [117].

where, λ is the wavelength, E is the kinetic energy, and E 0 is the rest-mass energy of the electron. Assuming that both V and B⊥ do not vary with zˆ , neglecting any stray magnetic and electric fields, we can simplify (9.3.17) as φ(x) = CE V (x)t(x) −

e 

 B⊥ (x)t(x)dx

(9.3.19)

Differentiating with respect to x, we get d dφ(x) e [V (x)t(x)] − B⊥ (x)t(x) = CE dx dx 

(9.3.20)

If the specimen is uniform in thickness and composition, the first term in (9.3.20) is negligible, and the in-plane induction is proportional to the phase gradient. However, in typical TEM specimens showing significant and rapid thickness variations, the mean inner potential contribution, V (x)t(x), can dominate and complicate the determination of the magnetic contribution. However, time rever-

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 617 sal operation of the electron beam—conceptually simple by physically inverting the specimen [40], but difficult to implement in practice—can be used to eliminate the effect of the mean inner potential. Currently, a phase sensitivity of π/50 can be routinely achieved in electron holography. Thus, magnetic induction on the spatial scale of 5 nm can be detected [41]. Further improvement in spatial resolution may require thicker specimens, larger intrinsic specimen magnetization, or longer data acquisition times.

9.4 Analytical Electron Microscopy (AEM) and Related Spectroscopies In addition to elastic scattering, which gives rise to the diffraction and imaging methods in a TEM (§9.3), inelastic interactions of the electron beam with the inner-shell, valence, or conduction electrons of the specimen provide information about the electronic structure and chemistry of the solid. The main inelastic processes (Fig. 9.4.1) can lead to intra-band, inter-band, and plasmon excitations, and the energy transferred to the specimen can be subsequently emitted in the form of secondary (§10.2.3) electrons, light, or CL (§10.6.2), or following innershell ionization, by the emission of characteristic X-rays (§2.3.2), or Auger (§2.3.3) electrons. Such inelastic scattering is significant; in fact, for elements with atomic number smaller than Cu, the total single electron inelastic scattering exceeds the total elastic scattering. The incident electron beam of energy, E 0 , loses energy characteristic of the process of excitation, and the energy it transfers to the specimen can be observed as an energy loss, E, of the incident electron reducing its kinetic energy to E 0 –E. A magnetic sector electron spectrometer (Fig. 9.4.8) mounted at the end of the electron column can record this electron energy loss-spectrum (EELS) and provide a wealth of information about the specimen; note that the underlying physics of EELS is similar to XAS (§3.9). In a STEM using a highly focused, and where possible, coherent probe, EELS is often combined with micro-diffraction (§9.4.5) or Z-contrast imaging (Fig. 9.3.11) to provide both crystallographic and electronic structure information at sub-nm spatial resolution. Note that in the EELS literature there is sometimes a tendency to separate the electronic structure from the crystal structure, but this is quite arbitrary as the two are strongly related in a solid. The addition of an energy dispersive X-ray spectrometry (EDXS) detector (§2.5.1.1), typically mounted on the electron optical column to detect and quantify the characteristic X-rays, completes the capabilities of such an AEM. An important advantage of EELS over EDXS is its collection efficiency (Fig. 9.4.2). Beam electrons inelastically scattered by inner-shell ionization processes are concentrated within small scattering angles in the forward direction

OUP CORRECTED – FINAL

618 Transmission and Analytical Electron Microscopy (b)

(a) E0

SE

CL

EV

AE

EF

EF

(i) (ii )

θE

X-ray

EV

E

+

(c)

(iii)

L1 L3

L2

E0–E K + Figure 9.4.1 Inelastic interactions of the electron beam with the specimen. (a) The related energy loss, E, results in scattering through very small angles, θ E , and is measured in electron energy-loss spectroscopy (EELS). (b) If E is small in magnitude it leads to (i) intra-band and (ii) inter-band transitions, as well as (iii) plasmon excitations that are collective oscillations of the valence band. If the excited electrons have energies greater than the work function, , they leave as secondary electrons (SE). Inter-band excited electrons can recombine by the emission of light (CL) or as a radiation-free phonon (not shown) generation. (c) At much larger energy losses, inner shell ionizations take place and the deexcitation process that follows produces either characteristic X-rays or Auger electrons (AE). Detecting the former is similar to EPMA (§2.5.2.2), and the latter is typically not detected in an analytical electron microscope. Note that for labeling purposes this figure shows oneelectron atomic orbitals even though in practice EELS measures differences in total energy between many-electron states in the solid. See Egerton (1996) for a more detailed discussion. and the EELS spectrometer, with a typical entrance aperture semiangle, β < 20 mrad, will collect them with efficiencies of the order of ∼50 %. On the other hand, characteristic X-ray quanta are emitted isotropically in all directions and only a small solid angle (∼0.1–1 sr) are detected by the EDXS detector, which results in very inefficient X-ray collection (or 0.8–8 % of all emitted X-rays) and requires long counting times to establish good statistics. In addition, since EELS observes the primary inelastic scattering event, unlike the secondary de-excitation process in EDXS, it can provide information on the wide range of possible inelastic interactions. These are introduced in the next section.

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 619 Beam Convergence Semi-Angle, α

EDXS Detector

X-ray Collection Semi-Angle, γ

X-Ray Take-Off Angle, φ

Incident Beam

Isotropic Characteristic X-Ray Distribution Specimen

EELS Collection Semi-Angle, β Scattering Angle, θ

EELS Spectrometer

Figure 9.4.2 Comparison of the spatial distribution and collection efficiencies of the beam electrons inelastically scattered in the forward direction that constitute the EELS signal, and the isotropically distributed characteristic X-ray signals, arising from deexcitation process, only a small fraction of which is collected by the EDXS detector. The figure also illustrates the experimental arrangements of EELS and EDXS in a TEM, and the scattering angle, θ, the electron-beam convergence semi-angle, α, and the EELS spectrometer acceptance semi-angle, β, as well as the X-ray take-off angle, φ, and the collection semi-angle, γ.

9.4.1 Inelastic Scattering and Spectroscopy Inelastic scattering processes arise from the interaction of the beam12 electrons with the electrons in the solid (Fig. 9.4.1). It conserves the total energy and momentum, and part of the kinetic energy of the incident beam electrons is converted to excitations of the electrons in the solid. Needless to say, the primary electrons lose the same energy and scatters through small angles in the forward direction (Fig. 9.4.2). For our discussion, we divide the inelastic scattering and the associated energy losses into three broad categories. At very low energy losses (20 meV–1 eV), we expect excitations of oscillations in molecules and gases, as well as phonon excitations in solids. Recall that the energy widths of the primary beam (Table 5.2.2) ranges from ∼1eV for

12 We also refer to beam electrons interchangeably as incident, primary, or fast electrons.

OUP CORRECTED – FINAL

620 Transmission and Analytical Electron Microscopy thermionic guns to 0.1–0.5 eV for field emission and Schottky sources. Hence, the observation of such very low energy-losses in a TEM requires monochromatization of the primary beam, using in-column energy filters (such as the -filter mentioned in §9.4.2.6). We shall not discuss this energy regime here, but further details can be found in the monograph by Ibach and Mills (1982) and the recent developments in high-energy resolution monochromators and spectrometers with sub-10 meV resolution [112] and sub-Å spatial resolution [113]. At slightly larger energy losses, E ∼ 1–50 eV, intra-band and inter-band excitations of the outer atomic electrons, and collective longitudinal oscillations of the valence or conduction electron bands, known as plasmons (Fig. 9.4.1b) with a broad maximum in energy-loss are observed. To first order, these discrete surface and volume plasmon losses, Ep , are proportional to the square root of the electron concentration, n, in the band, i.e. Ep ∝ n1/2 . Note that there is a clear distinction between the underlying physics of energy lost to plasmon excitations and to inner-shell ionization. In the former case, the initial state of the electrons in the band (see §3.2) occupies a range of energies but in the latter case they occupy a sharp energy level (Fig. 9.4.1c). As a result, unlike inner-shell ionization, the energy loss to collective excitation of the band electrons cannot be treated in terms of simple atom-like behavior but requires the treatment of the solid as a many electron system. This is difficult to do in quantum mechanics, and to overcome this difficulty it is common practice to describe plasmon excitations phenomenologically in terms of the dielectric behavior of the solid. Further details of this approach can be found in Raether (1980), Reimer (1995), and Egerton (1996). The cross-section for plasmon excitation for a primary electron of energy, E 0 , per unit solid angle, is given [42] per band-electron per unit volume as dσp θp 1 = d 2π a0 θ 2 + θp2

(9.4.1)

where a0 = 0.529 Å is the Bohr radius (§2.2.1), the characteristic plasmon scattering angle, θp = Ep /2E0 , the element of solid angle, d = 2π sin θdθ, and θ (0 < θ < π) is the scattering angle (Fig. 9.4.2) measured with respect to the trajectory of the incident beam. Typical values for aluminum for E0 = 100 keV are Ep ∼ 15 eV, θp = 0.15 mrad. This implies that the energy-lost electrons associated with plasmon scattering are highly forward peaked, and hence we can set d = 2π sin θdθ = 2π θ dθ , and integrate (9.4.1) to obtain the total plasmon cross-section:  σp =

dσp (θ) =

θp 2π a0

θ1 0

2π θ dθ θ 2 + θp2

(9.4.2)

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 621 Based on typical scattering angles in a TEM, we can safely set the upper limit of nc A the integration, θ1 = 0.2, and multiply by the factor ρN , where nc is the number 0 of conduction band electrons per atom, A is the atomic weight of the elements in the specimen, ρ is its density, and N 0 is the Avogadro number, to obtain a total plasmon scattering cross-section per atom/m2 as σp =



  nc A θp ln θp2 + (0.2)2 − ln θp2 2ρN0 a0

(9.4.3)

Since Ep ∝ n1/2 , in plasmon scattering the electron beam loses energy in quantized units that are detected as strong peaks in EELS. Moreover, the wavelength of the longitudinal plasmon oscillations is of the order of ∼100 interatomic spacings and the associated EELS signal is not localized. In other words, the spatial resolution of plasmon excitations in EELS measurements is determined by the plasmon wavelength (several nanometers) and not by the size of the focused probe. Last, but not least, inelastic scattering can lead to inner-shell (K, L, M, N, and O) ionization and the excitation of an electron either to an unoccupied state or to the continuum, well above the Fermi level. The EELS spectrum, dσ /dE, for the ionization of a core electron with binding energy, E c , will show an increase in intensity at energy loss E = Ec , followed by a long tail for E > Ec , classically referred to as a saw-tooth shaped curve. Other shapes such as a delayed maximum, “white lines,” etc., can also be observed (Fig. 9.4.5). In addition, for the energy region up to 50 eV above the onset of the edge, Ec < E < Ec + 50 eV, the energy loss near edge structure (ELNES), similar to NEXAFS (§3.9.2), which depends on the binding state of the core electron and its atomic environment, can be observed. Further, interference effects between the outgoing excited electron wave and the electron wave back-scattered from neighboring atoms, known as extended energy-loss fine structure (EXELFS), is observed in the form of long-range oscillations superimposed on the underlying edge shape. With analysis similar to EXAFS (§3.9.2), EXELFS gives information on the nearest neighbor distances and coordination number (Fig. 3.9.7). The cross-section, Q, for inner-shell ionization, attributed to Bethe [43], is given by Q=

  π e4 bs ns E0 6.51 × 10−20 bs ns log cs = log (cs U ) E0 Ec Ec UEc2

(9.4.4a)

where ns is the number of electrons in the specific shell, Ec is the ionization energy, bs and cs are constants appropriate for the specific shell, and U = E 0 /Ec is the overvoltage. The ionization cross-section, Q, varies with over-voltage (Fig. 9.4.3a) and the ionization probability is significant if U > 5. As the voltages used in TEM

OUP CORRECTED – FINAL

Q (a.u.)

622 Transmission and Analytical Electron Microscopy (a)

0

10

20

30

40 50 U

60

10–21

80 (b)

10–22 σ (m2)

Figure 9.4.3 Inelastic scattering cross-sections. (a) Variation of the ionization cross-section, Q, as a function of the overvoltage, U (= E 0 /Ec ). Above a critical value, U ≥ 5, the ionization is most probable. (b) The cross-sections for various inelastic scattering processes in aluminum as a function of acceleration voltage. The elastic crosssection is included for comparison.

70

Plasmon Elastic

10–23

L-Shell Ionization

10–24

K-Shell Ionization Secondary, ESE > 50 eV

10–25 Secondary, ESE < 50 eV 10–26 100

Adapted from Joy, Romig, and Goldstein (1986).

200

300

400

E0(keV)

are significantly high (E 0 > 100 keV), relativistic effects become important, and (9.4.4a) is modified as    

 m0 v2 π e4 bs ns log cs − log 1 − β 2 − β 2 Q = 2 m0 v 2Ec Ec 2

(9.4.4b)

where 

 1/2 v E0 −2 β = = 1− 1+ c 511

(9.4.5)

and E 0 is expressed in keV. The value of the constants, bs and cs , has been the subject of considerable discussion in the literature and are often adjusted for specific experiments. For low beam energies, E 0 < 20 keV, and small overvoltages, 4 ≤ U ≤ 25, for K shells, bs = 0.9 and cs = 1.05 [44]. For higher beam energies, 5.5 ≤ U ≤ 25, and higher atomic number elements, e.g. Ni, very different values of bs = 1.05 and cs = 0.51 are recommended [45]. Numerous other formulas to describe the inelastic cross-sections are also available [46, 47]. Figure 9.4.3b shows the cross-section for various inelastic scattering processes in Al as a function of the incident beam energy. Finally, a very important quantity that is involved in calculating scattering contrast is the ratio, γ , of the total inelastic scattering to the total elastic scattering cross-section:

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 623 γ (Z) =

σinelastic 20 ≈ σelastic Z

(9.4.6)

Figure 9.4.4 plots the inverse of this ratio, 1/γ (Z), is plotted in along with experimentally measured values; clearly σinelastic is much larger than σelastic for atoms with Z < 20 (compare γPt = 0.25 with γC = 3.3). In summary, inelastic scattering transfers energy from the incident electron beam to the specimen. This generates a variety of signals, principally EELS, which provides a wealth of information about the specimen. The de-excitation processes, following inner shell ionization, produce characteristic X-rays that can be detected using EDXS and quantified to obtain compositional information. These two methods, as implemented in a TEM, are discussed in the sections that follow.

9.4.2 Electron Energy-Loss Spectroscopy (EELS) in a TEM The experimental arrangement for EELS, including the important angles involved, is schematically illustrated in Figure 9.4.2. The incident beam of intensity, I 0 , and energy, E 0 , is inelastically scattered as it passes through the specimen into the analyzing spectrometer, which is mounted at the end of the electron-optical column (Fig. 9.1.3). In the process of inelastic scattering, it suffers an energy-loss, E, and emerges from the specimen with an energy, E = E 0 – E, and intensity distribution, I (E), or I (E). The spectrometer (Fig. 9.4.8) magnetically disperses I (E) such that all electrons emerging with a specific loss, E = E 0 – E, are brought to a focus at a specific point on the detector (§9.4.2.4). The entrance aperture to the spectrometer can be made narrow and positioned at any scattering angle, θ (Fig. 9.4.2). Then, the EELS spectrum collected as a function of the angle, θ, includes in addition to the energyloss, the momentum distribution of inelastically scattered electrons, and can be used to provide information on the complex dielectric function of the materials (see Raether, 1980). However, here we restrict our discussion to the case where the spectrometer integrates all electrons scattered up to a maximum collection angle, β (Fig. 9.4.2); by such angular integration the momentum information is sacrificed to gain the advantage of simple interpretation of the EELS spectrum in terms of the chemical and physical properties of the specimen. Now, the ratio I (E)/I 0 at any energy, E, will be proportional to the ratio of the inelastic scattering cross-section, σ inelastic (E, β), representing the probability of detecting an incident electron subject to a specific energy-loss, E, and scattered through a maximum angle, β, in the forward direction. The constant of proportionality, N, is equal to the number of atoms per unit area contributing to the observed scattering event. As we have seen in the last section, the inelastic scattering crosssection is related to specific physical interactions between the electron beam and the specimen, and if a standard specimen with a known concentration, N, is

4

Pt

3 1 γ 2 Ge Ni Cu Mn Ti

1

Sb

Al Si C

20

40 60 Z

80

Figure 9.4.4 Measured ratios of the elastic-to-inelastic scattering crosssection, 1/γ , as a function of atomic number, Z. Note the importance of inelastic scattering for Z < 20. Adapted from Reimer (1995).

OUP CORRECTED – FINAL

624 Transmission and Analytical Electron Microscopy used, the appropriate cross-section can be determined [48–50]. Alternatively, if the scattering cross-section is known, the concentration, N, can be determined from the measured intensity and quantitative information about the specimen composition, also known as microanalysis, can be obtained (§9.4.2.5). Figure 9.4.5 shows a representative EELS spectrum of an oxide material. It can be broadly divided into three energy regions: no-loss (NL), low-loss, and core-loss, and we shall now discuss them individually. 9.4.2.1 The No-Loss Region The no-loss region of the EELS spectrum (Fig. 9.4.5) contains its most visible feature, the zero-loss peak (ZLP), and includes three types of electrons: (i) those that are not scattered on passing through the specimen, (ii) elastically scattered electrons, and (iii) those that generate a phonon excitation with energy-loss, E < 1 eV. The energy width (full width at half maximum, FWHM) of the NL Low-Loss

Core-Loss

OK Hydrogenic Edge

Ba M

Ba N “White Lines”

Intensity (a.u.)

Delayed “Sleeping Whale” Profile

IA

Zero-Loss Peak Gd N

A

I0

Cu L

EA

Artificial Gain, 100×

Artificial Gain, 100×

AE–r

Plasmon IL 0

500 Energy-loss (eV)

1000

Figure 9.4.5 Electron energy-loss spectrum of GdBa2 Cu3 O7-δ , illustrating various parameters (I 0 , Zero loss peak intensity, I L , low loss intensity, E A , edge onset energy, A , integration energy window, I A , integrated intensity above background for element A), and the background model AE–r , described in the text. The figure also shows the plasmon loss and three types of edge shapes typically observed in EELS. Note the artificial gain applied in the display. Adapted from [51].

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 625 ZLP is a measure of the energy resolution of the experimental arrangement and is a function of both the energy spread of the source, including instabilities of the acceleration potential, and the energy resolution of the spectrometer. Taken together, typically the FWHM ∼ 0.2–1.5 eV, and depends on the type of source/gun as well as the stability of the monochromator used in the TEM. The angular distribution of the unscattered electrons—no interaction with the specimen—can be expected to be the same as that of the focused incident beam, i.e. a cone of angular range ∼1–5 mrad wide (Fig. 9.4.6a). The elastically scattered electrons, typically deflected by the positive nuclear charge, and given by the screened Rutherford cross-section (9.3.11), form a broad angular distribution (∼30 mrad) for amorphous materials (Fig. 9.3.2a). However, in crystalline materials, Bragg diffraction modifies this angular distribution with

1

(a) Incident Beam, E = 0

1 Crystalline Bragg Reflections

(b) Elastic Scattering, E = 0

Relative Intensity, I(E)/I0

Amorphous

1 (c) Phonon Scattering, E ~ 0.2 eV

1

(d) Plasmon Scattering, E ~ 20 eV

1 (e) Inner-Shell Ionization Scattering, E ~ 40–2000 eV

20 40 60 80 Scattering Angle, θ (mrad)

Figure 9.4.6 Angular distribution of (a) incident and unscattered electrons, (b) elastically scattered electrons (Rutherford cross-section) including the Bragg diffracted beams in crystalline materials, (c) phonon scattered electrons, which broaden the individual Bragg scattered beams, (d) plasmon scattered electrons, and (e) electrons scattered by inner-shell ionization with large energy losses. Adapted from Joy, Romig, and Goldstein (1986).

OUP CORRECTED – FINAL

626 Transmission and Analytical Electron Microscopy peaks corresponding to the Bragg angles (Fig. 9.4.6b). Note that the broad angular distribution of elastically scattered electrons impacts our ability to efficiently collect the inelastically scattered electrons, especially for thick specimens. Finally, the electrons that excite phonons, or vibration modes in the specimen, also contribute to the ZLP. They lose small amounts of energy (E ∼ 0.2 eV) but they have a broad angular distribution. Typically, in crystalline materials it is of the order of 5 mrad, and centered around each of the Bragg diffracted beams (Fig. 9.4.6c). 9.4.2.2 The Low-Loss Region This energy region of the EELS spectrum (Fig. 9.4.5) extends from the edge of the ZLP up to an energy loss, E < 40 eV, and includes the plasmon peak, especially in metals, as its principal feature. The integrated signal intensity, I L , in this region is about 5–10 % of the intensity, I 0 , of the ZLP. In practice, since most (∼80 %) of the electron beam passing through a thin specimen is either unscattered or elastically scattered, the total inelastically scattered electron intensity, I T , is essentially the sum of the no-loss and low-loss regions, i.e. I T = I 0 + I L . As we see shortly (9.4.8), this has implications in specimen thickness measurements. The energy losses in this region are a result of the direct electrostatic interactions of the electron beam with atomic electrons. It involves interactions or excitations with various bound states (Fig. 9.4.1b) including (i) intra-band, and (ii) inter-band transitions. However, the most prominent feature in this energy region is due to plasmon excitations, observed predominantly in metals, and attributed to the collective excitation of the delocalized electron “gas” or band. This excitation is not localized and occurs over a volume of several nanometers; further, the plasmon oscillations are rapidly damped with a typical lifetime (see §2.3.1) of 10–15 s. The energy loss, Ep , of the beam generates a plasmon of angular frequency, ωp , given by  Ep = ωp = 

ne2 ε0 m

1/2 (9.4.7)

where n is the free electron density in the band. The characteristic scattering angles, θp = Ep /2E0 , for plasmon scattering are rather small ( 50 eV, one observes edges corresponding to the interaction of the beam electrons with the deeply bound core electrons of the specimen. The intensity in this region is very much lower than in the no-loss and low-loss regions, and falls so rapidly with increasing energy-loss that an artificial gain is required to see the underlying features (Fig. 9.4.5). When the incident beam interacts with an atom in the specimen, a minimum energy corresponding to the binding energy, Ec , of the core electron is required for its ionization. Thus, the onset of the edge in the energy-loss spectrum corresponds to the ionization energy of the core electron. Moreover, as the binding energies of the core electrons are a unique function of the atomic number, the position of the onset of the edge in EELS can be used to unequivocally identify the elements that constitute the specimen and quantify its chemical composition. In addition, in the process of ejecting the core electron to the continuum, the beam electron can impart varying amounts of energy, up to a maximum of E 0 – Ec , to it. However, the probability of doing so decreases with increasing energy loss. As a result, the overall shape of an ideal core edge in EELS (see O K edge in Figure 9.4.5) is a sharp onset followed by a smooth decay, extending with decreasing probability to higher energy losses, and this tail often results in significant edge overlaps in multi-element specimens. Moreover, preceding edges and plasmon losses can impart considerable background to subsequent edges observed at higher energy losses. In practice, the background in EELS has been found to have the form A(E)–r , where A is a constant and the exponent, r, is usually in the range 3 < r < 5; both A and r are obtained by curve-fitting (Fig. 9.4.5; see [52, 53] and Egerton, 1986). The problem is compounded in thicker specimens due to plural scattering, but again they can be deconvoluted to obtain the single-scattered spectrum [54, 55].

OUP CORRECTED – FINAL

628 Transmission and Analytical Electron Microscopy The intensity of a core-edge in EELS varies as a function of the energy loss, E, and the scattering angle, θ . Hence, the cross-section is written in the form of a second differential:    1 dF d 2 σinelastic e4 = (9.4.9) d d (E) dE (4π ε0 )2 E0 E θ 2 + θE2 Here, the characteristic angle for inelastic scattering, θE = E/2E0 = 1.33 mrad for E = 532 eV (O K edge) and beam energy, E 0 = 200 keV, F is the generalized oscillator strength (GOS) that describes the response of an atom when a specific energy and momentum are supplied by the collision of a beam electron, and d F /dE, is the GOS per unit energy loss. Strictly speaking, to calculate (9.4.9) accurately, knowledge of the atomic wave functions and the band structure of the solid are required. For small scattering angles, θ, and energy-loss edge energy, Ec , the contribution of the GOS is a constant. Then, the angular distribution of  −1 inner-shell excitations is of the form θ 2 + θE2 , with a maximum value for the forward-scattered direction (θ = 0). Hence, it is safe to assume that the average inelastic scattering angle is of the order of ∼5 mrad, which is much smaller Z 1/3 than the characteristic angle, θ 0 (= λ2πa ), for elastic scattering (Fig. 9.4.6). In 0 addition, the cross-section shows a long tail and even though a majority of the inelastically scattered electrons are forward peaked, some of them are scattered at angles θ >> θ E . In practice, this results in not all the energy-lost electrons of the characteristic edge being detected, as finite collection angles (β < 50 mrad) are used to minimize electron-optical aberrations; in fact, both the shape of the spectrum and the intensities are affected by these finite apertures. Moreover, even though the inner-shell ionization signal will increase as the collection angle, β, is increased, eventually at some critical value, β ∼ (2θE )1/2 ∼ 50 mrad, the contribution from the GOS will tend to zero. Nevertheless, a substantial amount of the signal can be collected for β = 20 mrad. Finally, for large energy losses, E >> Eb , the contribution of the GOS is small in the forward direction (θ = 0), but increases to a maximum value (peak) at θ ∼ (E/E0 )1/2 ; known as the Bethe ridge, this effect is discussed in more advanced texts such as Egerton (1986) and Reimer (1993). Integrating (9.4.9) from E = E c to E = E 0 , gives the total cross-section for the inner-shell ionization, which is typically of the order of 10–24 m2 /atom for a light element, such as carbon at 100 keV. Compared to plasmons, this is at least two orders of magnitude weaker, and often requires the introduction of an artificial gain in displaying the core-loss spectrum (Fig. 9.4.5). Finally, a typical spectrometer collects the signal over a finite angular range (β) and energy window (A ) (Fig. 9.4.5). Therefore, in carrying out microanalysis (§9.4.2.5) only a fraction of the total cross-section, known as the partial cross-section, σinelastic (β, A ), is appropriate and used for quantification. When a small collection aperture (β ∼ 25 mrad) is used, the predominant transitions observed in an energy-loss spectrum are the ones governed by the

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 629 diploe selection rules (§2.3.1); in that sense, EELS spectra are very similar to X-ray absorption spectra (discussed earlier in §3.9). Standard spectroscopic nomenclature is used to label the core edges, i.e. K (1s), L1 (2s), L2 (2p1/2 ), L3 (2p3/2 ), M1 (3s), M2 (3p1/2 ), M3 (3p3/2 ), M4 (3d 3/2 ), M5 (3d 5/2 ), . . . , etc. In practice, only a few of these transitions can be recorded with ease and the relevant edges are those corresponding to an initial state of maximum l for a given n[56]. Hence, transitions originating in 2p, 3d, . . . initial states are an order of magnitude greater in intensity than those from 2s, 3s, 3p, . . . , etc. Three basic edge shapes (Fig. 9.4.5) are broadly observed; as the wave functions of the core electrons undergo little change upon forming a solid, a simple atomic model can predict their general shapes [57, 58]. They are: (a) “saw tooth” profile such as those for hydrogen-like wave functions for K-shell edges; experimentally observed edges, such as O K in Figure 9.4.5, conform to this general shape but with some additional fine structure near the edge onset; (b) a “sleeping whale” profile, as seen for the Ba N4,5 edge in Figure 9.4.5, or a delayed maximum observed ∼20 eV above the ionization edge, usually resulting from large centrifugal barriers due to the l’(l’ + 1) term in the radial Schrodinger equation, and commonly observed for the L2,3 edge for the third period elements Na – Ar; and (c) “white lines”, arising from distinct spin-orbit split levels and typically observed for the 3d transition metals (Fig. 3.9.3). As discussed in §3.9.2, L2,3 edges probe the d-symmetry portion of the final state wave functions, and can become large and narrow with sharp threshold peaks in a solid with high density of unoccupied d states. Similar effects can be observed in M edges as shown for Ba M4,5 (Fig. 9.4.5). The L3 /L2 ratio is often different from the statistical value of 2.0 based on initial state occupation (Fig. 9.4.7) and can be used to determine the oxidation state of the transition metal [59]. Superimposed on the broad edge shapes are the fine structures due to solidstate effects. At the edge threshold, one can measure a displacement of the onset, or “chemical shift”. In particular, positive chemical shifts are observed with increasing oxidation states (Fig. 9.4.7) because oxidation removes valence electrons, which leads to reduced screening of the nuclear field and a deepening of the potential well of the initial state. In EELS, the ionization edge threshold is a function of both the initial state as well as the position and nature of the vacant states around the Fermi level. The energy-loss near edge structure (ELNES), much like NEXAFS discussed in §3.9.2, and observed ∼3–50 eV above the ionization edge, can be interpreted in the first approximation using a simple one electron transition model between the initial state and a vacant final state as

I (E) = T (E) N(E)

(9.4.10)

where T (E), the transition probability, is a slowly varying function of energy loss, E, and N(E) is the density of final states. This simple model is reasonable for the interpretation of ELNES, provided N(E) is defined to include the

OUP CORRECTED – FINAL

630 Transmission and Analytical Electron Microscopy 2 eV

L3 L2 MnO2

Figure 9.4.7 The L3,2 edge for Mn2+ and Mn4+ illustrating both the chemical shift (2 eV) and the change in L3 /L2 ratio with the oxidation state. Adapted from [51].

MnO 624.8 eV

675.2 eV

following ideas: (a) dipole selection rules apply, i.e. N(E) is interpreted as a symmetry projected density of states clearly distinguishing the K and L edges; (b) core level states are highly localized, i.e. N(E) is a local density of states determined for that particular lattice site and reflecting the local symmetry; and (c) in reality, N(E), is a joint density of initial and final states and broadening, based on the lifetime of the core hole and the final states, is incorporated. For more advanced readers, the theory and analysis of unoccupied electron states in ELNES and XANES is discussed at length in Fuggle and Inglesfield (1992). Finally, long range oscillations of weak intensity similar to EXAFS (Fig. 3.9.6), superimposed on the high-energy tail of a core-loss ionization edge, gives information on the nearest neighbor distances and coordination number for that specific element in the specimen. Such extended energy-loss fine structure (EXELFS) data is analyzed in a manner very similar to EXAFS as discussed in §3.9.2. 9.4.2.4 The EELS Spectrometer and Signal Detection The magnetic spectrometer, also known as a magnetic prism because of its similarity to an optical prism that disperse white light into its frequency components, spatially disperses electrons according to their kinetic energy. In the spectrometer electrons traveling down the optic axis (z-direction) of the TEM (Fig. 9.4.8) encounter an orthogonal magnetic field of induction, B, along the y-direction. The electrons will experience a constant Lorentz force, FL = evB, in the region of the magnetic field and travel in a circular orbit of radius, R, which can be determined by simply balancing the Lorentz force with the centripetal force: R=

γ m0 v eB

(9.4.11)

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 631 

 2 −1/2

Here, γ = 1 − v2 /c is the relativistic correction factor, and m0 is the rest mass of the electron. The spectrometer is designed such that the electrons emerge from the magnetic region with a total angle of deflection, φ = 90◦ . Electrons that have suffered inelastic scattering in the specimen end up with different energies and velocities as they enter the spectrometer. Specifically, electrons undergoing larger energy losses will have smaller velocities, v, and smaller radius, R. Hence, they will leave the magnetic field region of the spectrometer with a larger deflection that is proportional to their energy loss. In addition, the magnetic sector behaves as a lens and brings all electrons of the same energy leaving the same object point to focus on the same point on the image plane of the spectrometer. Thus, electrons losing a specific amount of energy, E, compared to the primary beam of energy, E 0 , are brought to focus at a different point displaced by a distance, x = D. E, on the image plane, where D=

γ m0 v x 2R =2 = E0 E0 eB E

(9.4.12)

is called the dispersion of the spectrometer. For a microscope operating at E 0 = 100 keV, and an electron orbit radius, R = 0.2 m, we get a typical dispersion D = 4 μm/eV. These dispersed electrons can be collected, in principle, in two different ways: serial and parallel detection (Fig. 9.4.8). A narrow slit of width, ds , placed in the image plane of the spectrometer will allow only electrons of a specific and narrow energy loss to pass through to the detector. By ramping the magnetic field in small steps electrons of different energy (loss) can be detected serially by the detector. In this case, the minimum energy width, δEd , that can be detected by the slit would be δEd = ds /D, where D is the dispersion (9.4.12). Overall, the energy resolution of the spectrometer, δE, goes in quadrature with all its contributions, which include the energy spread, δE0 , of the source before the beam reaches the specimen, the energy resolution, δEso , of electron-optical components, and the spatial resolution of the electron detector given by the slit width, δEd . Thus (δE)2 = (δE0 )2 + (δEso )2 + (δEd )2

(9.4.13)

Alternatively, the entire spectrum can be detected in parallel using photodiode arrays (PDAs), as in the original design, or more recently replaced with CCD or direct electron detection cameras (Fig. 9.4.8b). If a PDA is used, the spatial resolution of the detector, defined by the interdiode spacing of the photodiode array, is ∼25 μm. For a typical spectrometer with a dispersion D = 4 μm/eV, calculated earlier, a PDA can only resolve ∼6 eV. To better the energy resolution, it is necessary to magnify the spectrum before projecting it onto the detector plane. A simple electron-optical lens can be used for magnification, but as we have seen before (§9.2.2), this will lead to magnification-dependent rotation of the image. To overcome this, a quadrupole lens system for magnification and

OUP CORRECTED – FINAL

632 Transmission and Analytical Electron Microscopy (a)

Object point

TEM

z

Figure 9.4.8 (a) Schematic layout of the EELS spectrometer for serial detection. Electrons leaving the object point from the specimen with the same energy are brought to focus at the same point on the image plane of the spectrometer. The dispersion, E/E, of the spectrometer depends on the magnetic field and is typically a few μm/eV. A slit of narrow width, ds , is used for energy selection. (b) The original design of a commercial parallel detection EELS system showing the pre-spectrometer focusing and alignment coils (QX, QY, SX, SY), quadrupole lens array (Q1–Q4), and position sensitive detector, which is a YAG scintillator couple to a photodiode array (PDA). Nowadays, the PDA is replaced with CCD cameras or direct electron detection cameras.

R Image plane Magnet

ds E0 –E

B

Detector

x E0 Slit (b)

QX QY SX SY ALIGN

Magnet

electrically isolated drift-tube

Adapted from Egerton (1986).

beam trap aperture

photodiode array

Q1 Q2 Q3 Q4 and transverse deflector

thermoelectric cooler

YAG

fiber-optic window

rotation-free focusing of the spectrum on the detector plane is used [60]. Note that in both cases, good energy resolution requires that an electron-beam crossover of small diameter be placed at the spectrometer object plane. This is best accomplished by positioning either a low-magnification image (image coupling, or seeing a diffraction pattern on the viewing screen, as the spectrometer looks at final projector cross-over) or the central diffracted beam (diffraction coupling, or seeing an image on the viewing screen) at the object point of the spectrometer. Finally, it is important to mention that serial spectrometers are no longer available commercially. 9.4.2.5

Microanalysis Using Inner-Shell Ionization Edges

In addition to the analysis of near-edge fine structure (ELNES) to obtain bonding and chemical information, EELS is used for quantitative microanalysis, especially for light elements (Z < 11), where it has significant advantage in the TEM over the alternative energy-dispersive X-ray spectroscopy (EDXS) method (§9.4.3).

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 633

Intensity

However, the detection of hydrogen in a solid is complicated because other lowloss features (ZLP and plasmons) often obscure the H1s edge. Moreover, in metallic hydrides, the 1s electron is often incorporated in the conduction band, resulting in a shift of the host Fermi level. Nevertheless, shifts in the plasmon energy observed in hydrides have been interpreted in terms of the composition in a variety of metal hydride systems [61, 62]. However, such interpretation is difficult as it involves understanding the modification of the band structure due to the addition of hydrogen, and often leads to detection limits one order of magnitude worse than the simple EELS microanalysis (described later) using core edges for Z > 3. For a specimen that is thin enough to avoid plural scattering, typically 20–50 nm in thickness, the quantification procedure is straightforward and involves measurement of the area of the appropriate ionization edge, after background subtraction. There are a number of methods available in the literature (Egerton, 1996; Williams and Carter, 1996) to obtain such an integrated intensity from the energy-loss spectrum, but we present the two-area method of background fitting to the power law form AE –r , and determining the constants A and r. Consider the ideal spectrum (Fig. 9.4.9) showing the energy regions and integrated intensities required for background fitting and quantification. The background region, E 1 – E 2 , over which the fitting is desired, is divided into two equal halves with measured integrated intensities, I 1 , and I 2 , as shown. Then we can obtain the power law constants as

IA I1 I2 I3 E1

E2E3

E4 Energy loss (eV)

AE –r

Figure 9.4.9 A portion of an ideal energy-loss spectrum showing the regions needed for fitting the background, its power-law (AE–r ) extrapolation under the edges, and the integrated peak intensity.

OUP CORRECTED – FINAL

634 Transmission and Analytical Electron Microscopy r=

2 log (I1 /I2 ) log (E2 /E1 )

(9.4.14)

and A=

(I1 + I2 ) (1 − r) E21−r − E11−r

(9.4.15)

Once these constants are known, the background can be extended into the region, A = E 4 – E 3 , under the edge, and the required integrated intensity, I A (β, A ) for the element A, can be obtained from the gross integrated intensity, I 3 , as IA (β, A ) = I3 − A

E41−r − E31−r 1−r

(9.4.16)

The area, I A , or the counts for an element A, is a product of the incident current, J 0 , the number of atoms, N A , per unit volume of the specimen, the thickness, t, of the specimen, and the total ionization cross-section, σ A , for the excitation of the appropriate inner-shell by the incident electrons. However, the entrance aperture to the spectrometer limits collection of inelastically scattered electrons to angles less than β, and hence only a fraction, I A (β), of the core-loss signal is measured. Moreover, in most microanalysis situations with multi-element specimens, to avoid edge overlap, the intensity, I A (β, A ), over a limited energy-loss window, A , following the ionization edge, E A , is measured (Fig. 9.4.9). Then the absolute concentration, N A , of the element A is NA =

IA (β, A ) GJ0 σA (β, A ) t

(9.4.17)

where G is any artificial gain incorporated in the spectrum, and σA (β, A ) is the partial ionization cross-section corresponding to the scattering angle, 2β, and inner-shell losses between energies E A and E A +A . If plural scattering can be avoided or corrected for by appropriate deconvolution methods, J 0 can be replaced with the area under the zero-loss peak, I 0 (Fig. 9.4.5). Often, only relative abundances of two elements, A and B, are of interest. Then, (9.4.17) gives IA (β, A ) σB (β, B ) NA = NB IB (β, B ) σA (β, A )

(9.4.18)

where it is assumed that the data for the two edges are measured under identical experimental conditions of illumination, specimen thickness and scattering angles, and the artificial gains, GB and GA are the same. Then, the accuracy in the analysis is largely determined by errors arising from the removal of the background using

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 635 the AE–r form, and the correctness of the partial ionization cross-section, either measured or calculated, used in the analysis. Two methods of calculating partial ionization cross-sections are generally in use. An approximate, but easily programmable model, SIGMAK, for K-shell edges based on hydrogen-like wave functions and scaled to account for the nuclear charge, along with a screening constant independent of Z, has shown good agreement with experimental measurements [63]. For L-shells, the equivalent SIGMAL [64], uses an additional empirical factor to match experimental data as the simple treatment for screening is inaccurate [65]. Both SIGMAK and SIGMAL can also be obtained by a parametric approximation by writing the partial ionization cross-section for the element A, with edge onset at E A (eV), and energy window, A (eV) as σA (β, A ) = σ ·f (β) ·g (A )

(9.4.19)

where the saturation cross-section, σ , is given by

σ =

F ln

0.055E0 EA

E0 ·EA

 m2 /atom

(9.4.20)

and E 0 is the incident beam energy in eV, and the constant, F = 1.60 × 10–17 m2 /atom for K-edges, and F = 4.51 × 10–17 m2 /atom for L-edges. The characteristic inelastic angle, θEA , for the average energy-loss θEA =

EA + A /2 2E0

(9.4.21)

can be used to calculate the second term, f (β), in (9.4.19) as

2   log 1 + β/θEA   f (β) = log 2/θEA

(9.4.22)

where, β is the spectrometer acceptance semi-angle in radians. The last term, g(A ), in (9.4.19) has the form

g (A ) = 1 −

EA EA + A

s−1 (9.4.23)

and the exponent s is given by

 s = H − 0.334 ln β/θEA with the constant H = 5.31 for K-edges, and H = 4.92 for L-edges.

(9.4.24)

OUP CORRECTED – FINAL

636 Transmission and Analytical Electron Microscopy The cross-sections can also be calculated more accurately using Hartree– Slater (HS) wave-functions [58, 66], assuming that the element is in atomic form, and neglecting solid-state or exciton effects. In the calculations of partial ionization cross-sections both methods agree to within 5 % for K-shell and 10% for L-shell edges. Alternatively, we can measure the partial ionization cross-sections using standard specimens and two systematic measurements, one for K- and L-edges [67], and the other for M-edges [68], and both are available in the literature. However, some of the experimentally measured cross-sections show large variations [65], and some have been collected with large collections angles (β∼50 mrad) that may be susceptible to errors due to lens aberration effects [69]. Finally, using the integrated edge-to-background ratios, minimum detectable mass (MDM) has been defined for EELS microanalysis; typically, using current technology, MDM ∼ 10–19 g. Further details can be found in Joy, Romig, and Goldstein (1986), Egerton (1996), and Williams and Carter (1996). Example 9.4.1: An unknown compound shows two edges at (A) 188 eV and (B) 284 eV in an energy loss spectrum. Following Figure 9.4.9, for edge A, the energies are E 1 = 163 eV, E 2 =186 eV, E 3 =188 eV, and E 4 = 258 eV. The corresponding intensities are I 1 = 36829, I 2 = 28226, and I 3 = 107112 counts. Similarly, for edge B, the energies are E 1 = 240 eV, E 2 = 280 eV, E 3 =284 eV, and E 4 = 354 eV, and the intensities are I 1 = 26189, I 2 = 20132, and I 3 = 53273 counts.Identify the elemental edges A and B, and determine their chemical ratio. Make any appropriate assumptions. Solution: We first make the reasonable assumption that these spectra were taken in a TEM with E 0 = 100 keV, and with a typical spectrometer acceptance semi-angle β = 3 mrad. The integration window  = E 4 − E 3 = 70 eV is the same for both edges. Looking at the binding energies for the core electrons (Table 2.2.2), we can determine that A is the Boron (B) K-edge, and B is the Carbon (C) K-edge. We can calculate the parametrized SIGMAK partial ionization crosssections using (9.4.19–9.4.24). Thus σKB (β = 3 mrad,  = 70 eV) = 5.84 × 10−25 m2 /atom σKC (β = 3 mrad,  = 70 eV) = 8.33 × 10−26 m2 /atom Next, we have to analyze the spectra and determine the integrated intensities. In order to do this, we have to do the power-law fitting of the backgrounds. Using (9.4.14) and (9.4.15) we determine, A and r for the two edges. For the B K -edge and C K-edge we get r B = 4.03, AB = 3.017 × 1012 and r C = 3.412, AC = 1.982 × 1011 . Then, we can extrapolate the background under the respective edges, and using (9.4.16) determine the integrated intensities in each edge as

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 637

IB (β = 3 mrad,  = 70 eV) = 28128 IC (β = 3 mrad,  = 70 eV) = 12034 We can now determine the relative ratios of the two elements using (9.4.18) as IB (β, A ) σCK (β, B ) NB 28128 × 8.33 × 10−26 = = 0.3334 = K NC IC (β, B ) σB (β, A ) 12034 × 5.84 × 10−25 In other words, this compound is BC3 . In fact, such a compound was newly synthesized and analyzed by EELS many years ago [114].

9.4.2.6

Energy Filtered Imaging

In an electron microscope, we can combine the chemical information in EELS with the spatial information of the microstructure at a resolution better than 1 nm [70, 71]. Since this technique is based on EELS, energy filtered transmission electron microscopy (EFTEM) can be a powerful method to study light elements (3 < Z < 11), and as such it is particularly applicable to the study of biological materials where the elements, C, N, and O are predominant. The technique uses an imaging filter attached to a TEM that makes it possible to create images only with electrons that have lost a specific energy. The simplest application of EFTEM is to filter out unwanted inelastically scattered electrons to improve contrast and interpretability of both TEM images and diffraction patterns (Fig. 9.4.10). It can also be used to selectively image the specimen with inelastically scattered electrons that have lost a specific energy; we now discuss this further. The imaging filter focuses the image onto an energy-dispersive plane, where an energy “window” selects only those electrons that have lost a specific energy, E ± δE/2, where E = E0 − E is the selected energy, and δE is the energy width of the electrons used in imaging. There are two commercially designed image filters available: the Gatan image filter (GIF) [73], mounted at the end of the electron-optical column, and the in-column Zeiss omega filter ( -filter) [74]. The GIF is similar to the EELS spectrometer with parallel detection (Fig. 9.4.8b), but with the addition of a set of quadrupole and sextupole lenses to focus the image onto a CCD camera. The -filter is mounted in-column, and includes a set of four magnetic sectors that bend the electron beam in the shape of (hence the name) to form an energy dispersed diffraction pattern at the exit plane of the filter along the original optic axis of the TEM. Then, an aperture can select the required energy range for the image. From an information point of view, the EELS spectrum can be considered as an additional energy dimension to the image. Thus, we have a three-dimensional data set for the specimen, combining a conventional spatial image I (x,y) with the spectroscopic information I (E), at each point, (x,y), as shown schematically in

OUP CORRECTED – FINAL

638 Transmission and Analytical Electron Microscopy (i)

(ii)

2

(a)

(b)

(c)

(d)

3 1

Figure 9.4.10 (i) Comparison of electron diffraction patterns from a thin crystal of F41 (P21 , a = 51.8 Å, b =36.5 Å, c = 118.7 Å, β = 90.8◦ ) without (left) and with (right) energy filtering. The background intensity near the beam stop is almost 100 times and overall, 5.5 times more on the left. (ii) Microtwins in an epitaxially grown film of Ag on NaCl with (a) unfiltered and (b) zero-loss filtered bright field images, and (c) unfiltered and (d) zero-loss filtered dark field images obtained by shifting the objective aperture. (i) Adapted from [72]. (ii) Adapted from Reimer (1995).

Figure 9.4.11a. Note that this approach can also be used to obtain data sets in higher dimensions, for example, by adding the time dimension such as in fourdimensional electron microscopy; see §9.5.5 and Zewail (2008). The acquisition of such a comprehensive three-dimensional data set can be carried out in two different ways, depending on whether a conventional TEM or a dedicated STEM is used. In the former, EFTEM, images covering a specific energy-loss range selected by the energy-selecting slit are collected in parallel by the CCD camera to form two-dimensional images (Fig. 9.4.11b); the complete three-dimensional data set is obtained by acquiring a sequence of images using different energy windows (horizontal sectioning of the data cube). In the latter, STEM-EELS (Fig. 9.4.11c), the energy-selecting slit is removed and the entire energy loss spectrum is collected in parallel at each pixel (vertical sectioning of the data cube). The EFTEM typically provides images of larger areas (105 – 107 pixels) compared to STEM-EELS, which contains about 104 (100 × 100) pixels and takes about 103 s of acquisition time (= 10–1 s dwell time/pixel). Figure 9.4.12 shows the complementarity of the two methods; note that for a given dose the signal is always higher for STEM-EELS compared to EFTEM, an important factor in imaging biological materials that are highly dose sensitive. To obtain quantitative images representative of the chemical composition of the specimen, the three-window elemental mapping method is often used. Figure 9.4.13 shows a schematic of the method. Three images are acquired, two pre-edge to

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 639 (a)

(b) EFTEM STEM–pixel by pixel

(c) STEM–EELS Source

Source

CL

x Intensity Zero-loss

y

Specimen

Scan coils OL

Plasmon Objective aperture

E

PL Core-losses

Energy– selecting slit

CCD detector

E

Magnetic prism

Quadrupole sextupole lenses y

y x

E

x

E

Figure 9.4.11 (a) Schematic of the 3D data set cube obtained by spectroscopic imaging using energy-loss spectroscopy; x and y are spatial coordinates, and E = E 0 − E is the energy axis. Comparison of compositional imaging in (b) a conventional energy-filtered transmission electron microscope (EFTEM) fitted with a post-column energy filter (an in-column energy filter will also work), and energy-selecting slit. (c) In a scanning transmission electron microscope (STEM-EELS), the slit is removed and the EELS spectrum is collected in its entirety at each pixel. In these two cases the 3D data sets (x, y, E) are built up differently either by (b) horizontal or (c) vertical sectioning of the (x, y, E) cube as shown. Adapted from [77].

estimate the background under the edge by extrapolation, and one post-edge. By subtracting the extrapolated background pixel by pixel from the post-edge image, the chemical map is formed. An alternative to this method is the jump ratio method, which divides the post-edge image by the pre-edge image. Such a jump ratio image provides qualitative images of changes in the spectrum at the edge of interest, but it is not quantitative (see §1.4.4. for basics of digital imaging). Figure 9.4.14 shows EFTEM images of a thin film standard with varying thickness of Mn layers, sandwiched between PdMn layers, using both the threewindow and jump ratio methods. It is clear that EFTEM can provide sub-nm resolution in these images. Further information on energy-filtered imaging can be found in review articles [75, 76], and in the dedicated monograph on the subject edited by Reimer (1995).

OUP CORRECTED – FINAL

640 Transmission and Analytical Electron Microscopy Dose for 3 nm Pixels (electrons/nm2) 104

106

t = 10 s I = 10 nA

108

107

109

1010

10

t = 100 s I = 100 nA

105

1

104

t = 1000 s I = 10 nA

t = 100 s I = 1 nA

103

0.1 STEM–EELS

102 3 10

104

105 106 107 108 109 Number of electrons/pixel

Pre-edge 1

1010 1011

Post-edge

Elemental Map

E

E1

E2

Excitation specific signal

Post-edge

Pre-edge 2

I

Pre-edge 2

Pre-edge 1

Adapted from [75].

106

EFTEM

Adapted from [77].

Figure 9.4.13 Illustration of the three-window technique for a multilayer specimen of Prx Ca1-x MnO3 , with x = 1, 0, 0.25, 0.5, 0.75, 1, respectively, for each layer grown on a substrate of SrTiO3 . Two preedge images are used to extrapolate the background, following a fit to the AE–r model. The post-edge image is acquired and the extrapolated background image is subtracted to give the Pr elemental map image, in agreement with the Pr content in the layers.

105

107 Number of pixels

Figure 9.4.12 Comparison of EFTEM and STEM-EELS imaging modes showing the availability of total number of pixels and the number of electrons per pixel. Different beam currents, I, and acquisition times, t, are shown by diagonal lines. Assuming 3-nm width pixels, the total image width is shown on the right, and the total electron dose is shown on top. Note that for a given dose, the signal is always higher for STEM-EELS.

103

Image width for 3 nm Pixels (μm)

102 108

E3

AE –r

E

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 641 TEM bright field image

(a)

Mn L elemental map

(d)

(c)

(b)

Relative thickness map (t/λ)

Mn L jump ratio image

Pd M elemental map

(e)

Pd M jump ratio image

(f)

20 nm

Figure 9.4.14 EFTEM images of a Mnt (PdMn)3.7 multilayer specimen with t = 2.6, 2.2, 1.8,1.4, 1.0, and 0.47 nm showing (a) BF image, (b) Mn three-window elemental map, (c) Mn jump-ratio map, (d) relative thickness (t/λ) map, (e) Pd three-window elemental map (Pd M4,5 ), and (f) Pd jump-ratio map. Note that even the thinnest Mn layers are clearly resolved, confirming sub-nm spatial resolution. Adapted from [71].

9.4.3 Quantitative Microanalysis with Energy-Dispersive X-Ray Spectrometry Following inelastic scattering and an inner-shell ionization event, the specimen responds by one of two possible de-excitation processes involving either the emission of a characteristic X-ray photon or an Auger electron (§2.3). As mentioned, characteristic X-rays can be detected in a TEM by incorporating an energy dispersive X-ray detector (§2.5.1.2), positioned at a well-defined take-off angle,

OUP CORRECTED – FINAL

642 Transmission and Analytical Electron Microscopy φ, to collect a small fraction of the total X-ray emission (Fig. 9.4.2). The X-ray spectrum detected by the EDXS detector (Fig. 9.4.15) can be used to measure the characteristic X-ray intensities, above the background signal, I A , I B , . . . , of the various elements, A, B, . . . , in the specimen and quantify them in terms of the chemical composition, i.e. mass concentration, C A , C B , . . . , using either a simple ratio method or by using appropriate thin film standards. We begin by making the simple assumption that in an electron-transparent, thin foil TEM specimen, very few incident electrons are back-scattered, and most of the beam electrons follow a trajectory roughly equal to the thickness, t, of the specimen at the point of analysis (note that in a typical wedge-shaped TEM specimen, t can vary from point to point). We also assume that, on average, the beam electrons lose very little energy (5 eV/nm) on traversing through the specimen. Then, for a specimen satisfying the thin film criteria (above), for any element A, the characteristic X-ray intensity, I A , measured over a specific time interval is given by IA = κCA ωA QA

t ηA AA

(9.4.25)

where κ is a proportionality constant, dependent on instrumental parameters including the incident beam intensity and the detector geometry, ω is the fluo-

O

Mg Carbon Coating

0.00

Element and Line

Si

Ca

k-factor

Atomic %

O Kα

1.70

60.77

Mg Kα

0.95

7.76

Al Kα

0.90

0.81

Si Kα

1.00

19.27

S Kα

1.07

0.26

Ca Kα

1.04

6.46

Fe Kα

1.40

4.66

Fe Cu Grid

9.31 keV

Figure 9.4.15 EDXS spectrum and microanalysis using experimentally measured k-factors of a thin film glass standard (K411) obtained from NIST. A 200 kV AEM, with a LaB6 source and an ultra-thin window detector were used [79].

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 643 rescent yield for characteristic X-rays (Fig. 1.4.5), Q is the cross-section for a particular inner-shell (K, L, M, . . . , etc.) ionization given by (9.4.4a) that varies with the incident beam energy, E 0 , and the binding energy, Ec, of the core electron, through the overvoltage, U (=E 0/ Ec ), A is the atomic weight and η is the efficiency of detection for characteristic X-rays of a specific energy; in all cases, the subscript refers to a specific element A. As mentioned (§5.1.1), the EDXS detector has various layers—the Be or ultra-thin window (if used) layer, the Au surface layer, and the Si dead layer—each of different thickness where the characteristic X-rays can be absorbed. Then, the efficiency, η, of detection is given by the sum of the absorption contribution, (2.4.2), of these individual layers as 

     Au  Si μ Be(UTW) μ μ ηA = exp − ρBe(UTW) XBe(UTW) − ρAu XAu − ρSi XSi ρ A ρ A ρ A     μ Si 1 − exp (9.4.26) ρSi YSi ρ A where the thickness of the layers, XBe(UTW) , XAu , and XSi , as well as the thickness of the active silicon layer in the detecting crystal, YSi , vary from detector to detector, and are not easily available. However, since the intensities I A , I B , . . . , for different elements are obtained simultaneously under the same conditions in a given EDXS spectrum, a simpler method of ratios [78] can be applied. Thus, the ratio of intensities:

 Qω ηA  C  A IA A (9.4.27) = A Qω IB CB η B A B

is proportional to the mass concentration ratio, and where the proportionality constant, κ, and thickness t, cancel out. This is generally written as CA IA = kAB CB IB

(9.4.28)

and kAB is referred to as the Cliff–Lorimer k-factors, or simply as the k-factors, for elements A and B. In a binary alloy, we have the additional relationship, CA + CB = 1, which allows us to solve for the concentrations. Similarly, in a ternary system, we have IA CA = kAB CB IB CC IC = kCB CB IB CA + CB + CC = 1

(9.4.29)

OUP CORRECTED – FINAL

644 Transmission and Analytical Electron Microscopy which can be solved for the concentrations provided kAB and kCB are known. There are two ways to determine the kAB factors: either by calculating the cross-sections and using (9.4.27), or using a standard specimen with known concentrations, C A , C B . . . , by measuring the intensities, I A , I B . . . , and determining kAB . . . from (9.4.28). It is common practice to standardize the kAB factors with respect to Si. Then kAB =

13 https://www.nist.gov/publications/ status-microanalysis-standards-nationalinstitute-standards-and-technology-nist

kASi kA = kBSi kB

(9.4.30)

where, in the end, the subscript, Si, is understood and not explicitly included. The k-factors for various inner shell excitations have been measured experimentally; Figure 9.4.16 shows a careful set of measurements for a 200 kV instrument, using an ultra-thin window detector. Using these k-factors and (9.4.20) modified for more elements, the spectrum (Fig. 9.4.15) for the glass standard obtained from NIST,13 was analyzed with good agreement. Note that most commercial detectors come with a software package for data acquisition and analysis that includes the ability to calculate the ionization cross-sections and k-factors using in-built parameters for the detector. This is adequate for most common microanalysis applications; however, to make the most accurate analysis, there is no substitute to the measurement of k-factors for a particular EDXS detector, using standards of known composition. In addition, to correct for the absorption of the characteristic X-rays in the specimen (§2.5.2.3) accurate measurement of the thickness, t, at each analysis point is necessary; if the standard is a crystalline material, the CBED method (Fig. 8.6.11) can be used to measure the local

Detector parameters 8

Layer

Thickness

UTW Parylene window

0.10 μm

UTW Au coating

0.15 μm

Au contact layer

20.0 nm

Si dead layer

0.125 μm

kASi-factors

6

Figure 9.4.16 Experimentally measured k-factors for the K lines for all the principal elements up to Cu, using an UTW detector. The inset shows the measured thicknesses of the principal layers of the detector required for accurate microanalysis. Adapted from [79].

4

2

0 B

C

0.185

N

O

Mg AI

1.25

Si

Ca

Ti

3.89 Energy (keV)

Mn Fe

Ni Cu Ga As

6.93

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 645 specimen thickness, t. Details of such accurate k-factor measurements, including the corrections for local specimen thickness, are given in [79]. Recently, a new quantitative thin-film X-ray analysis procedure, termed the ζ -factor method and which overcomes the two principal limitations of the Cliff–Lorimer method—(1) use of pure-element rather than multi-element thin-specimen standards, and (2) built in X-ray absorption correction with simultaneous thickness measurements— has been proposed. Details can be found in [115]. The X-ray counting statistics follow Gaussian behavior; hence the standard √ N, where N is the number of counts in the peak above deviation, σ = √ background. For 3σ level of confidence, the error in the measurement is 3 N with a relative error of 3N –1/2 . Thus, for the ratio method of microanalysis, the relative error in the measurement of the ratio C A /C B , is the sum of the errors of I A , I B , and kAB . In practice, accumulating ∼10,000 counts in each peak will guarantee a relative error of 3% in the measured concentration. Another criterion important in microanalysis is the minimum mass fraction (MMF) that can be detected. Statistically, it can be expressed as MMF =

  −1/2 P ×P×τ B

(9.4.31)

where P is the elemental counting rate, P/B is the peak to background ratio, and τ is the counting time. To improve MMF, all three terms should be maximized. To improve P, it is best to increase the current density in the electron beam/probe, which is limited by contamination rate, beam sensitivity of the specimen (§5.3.2), and the stability of the stage. Further, for a given detector location (take-off angle, φ), the specimen should be tilted such that the X-ray signals are maximized and minimally absorbed in the specimen. And, where possible, the highest beam energy available should be used (why?). For more practical details on EDXS microanalysis, interested readers should consult Williams and Carter, Volume IV (1996).

Example 9.4.2: (a) Using the STEM probe described in Example 9.2.7, we carry out an EDXS microanalysis of a thin film, 30 nm thick, of iron with Cr impurity. The beam rapidly contaminates the specimen in about 10 seconds. Assume that the background is 2% of the Fe peak, and the Cr peak can be detected if its intensity is at least five times the statistical variation of the background. If the minimum detectable mass of Cr is 1 at %, what should be the minimum number of counts of Fe Kα per second to make this analysis possible?

OUP CORRECTED – FINAL

646 Transmission and Analytical Electron Microscopy (b) How does the minimum detectable concentration of Cr change if the probe diameter is increased to 100 nm? (c) What will happen if the specimen thickness is doubled to 60 nm, and the probe diameter is kept at 10 nm? Solution: (a) Assume that the we detect N counts of Fe Kα per second. Then, for 10 seconds, the background is 0.02 × 10 × N = 0.2 N counts, with a statistical variation of (0.2 N)1/2 . For Cr to be detected, the intensity of Cr Kα, should be at least 5 × (0.2N)1/2 counts. We assume that the crosssections for Fe and Cr are the same. Then, for a minimum detectability of 1 1/2 1 at % we have 5×(0.2N) = 100 . 10N Solving for N we get N = 500 counts of Fe Kα per second. (b) When the probe diameter is increased by a factor of 10, the total counts of iron will increase by a factor of 100. Thus in 10 seconds, before contamination stops the experiment, we will count a total of 500 × 10 sec × 100 = 500,000 counts of Fe Kα. This will have 0.02 × 500,000 counts in the background, with a statistical variation of (0.02 × 500,000)1/2 = 100. Thus, for Cr to be detected, we should have a minimum of 5 × 100 = 500 counts. Then, the minimum detectability of Cr would be 500/500000 = 0.1 at % of Cr. (c) If the specimen thickness is doubled, assuming there is no multiple scattering, the detection limit will improve by a factor of 21/2 , when compared to (a). Thus the minimum detectability of Cr would be 1/21/2 = 0.707 at %.

Example 9.4.3: Quantitative analysis combining EDXS and EELS: There are significant difficulties in the analysis of light elements, such as B, in a heavy rare-earth matrix. The microstructure of a Fe–Nd–B alloy, shown in the following figure (a), contains an Fe–Nd matrix and a second phase (RE) containing B. The quantitative analysis of B by EDXS is difficult, and the analysis of Nd by EELS is difficult because the cross-section for the M-edge is not readily available. A 100 kV TEM equipped with an EDXS and EELS spectrometer is used for all analysis. We use 100 eV energy windows for the analysis of all elements, and a collection half-angle β = 13.5◦ for EELS analysis.

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 647

(b)

Fe

Iron

(c)

0.1 μm Matrix Nd

RE 1

(d)

(a)

170

Fe Cu-grid

3 5 7 Energy (keV)

9

Boron

210 230 Energy-loss (eV)

(a) An EDXS spectrum as shown in Figure (b) is acquired from the matrix of known composition. For this particular detector the k-factors, kFeSi = 12.6 and kNdSi = 1.8, and the measured intensities for the peaks are I Fe = 3,150 and I Nd = 450 counts, what is the composition of the matrix phase? (b) We collect an EELS spectrum from the matrix and observe an Fe L-edge and a Nd M-edge. We calculate a cross section σ Fe-L = 2.75 × 10–21 cm2 /atom. The measured intensities in the edges, similar to the ones shown in (c) and (e) are I Fe-L = 8.6 × 106 and I Nd-M = 4.88 × 106 counts, respectively. What is the cross-section for the M-edge of Nd? (c) We can also calculate the B K-edge cross-section as σ B-K = 9.85 × 10–20 cm2 /atom. We now have all the required cross-sections for the EELS analysis of the unknown Fe–Nd–B phase. If the measured intensities are I B-K = 9.998 × 105 , I Fe-L = 4.189 × 103 and I Nd-M = 1.30 × 104 counts, respectively, what is its composition?. Solution: (a) We determine the k-factor kFeNd = kFeSi /kNdSi = 12.6/1.8 = 7.0. From (2.5.7), for the unknown atomic concentrations, CFe and CNd , we get I Fe /I Nd = kFeNd (CFe /CNd ) = 3150/450 = 7 (CFe /CNd ). Thus CFe /CNd = 1, and assuming that this is a pure binary alloy, CFe + CNd = 1. Hence, CFe = CNd = 0.5. Thus, the matrix phase is an equiatomic alloy FeNd.

660

720 780 Energy-loss (eV)

(e)

920

Neodynium

960

1000 1040

Energy-loss (eV)

OUP CORRECTED – FINAL

648 Transmission and Analytical Electron Microscopy (b) We apply (9.4.18) to calculate the unknown cross-section for the Nd M-edge. Thus IFe−L σNd−M INd−M σFe−L . Hence, σ Nd-M 1.56 × 10−21 cm2 /atom. NFe NNd

=1=

=

INd−M IFe−L σFe−L

=

4.88 −21 8.6 2.75 × 10

=

(c) For the unknown, we apply (9.4.18) for pairs of elements as follows NFe NNd

=

IFe−L σNd−M INd−M σFe−L

Similarly, and

NNd NB

NFe NB

=

=

=

4.189 x 103 1.56×10−21 1.30 x 104 2.75×10−21

IFe−L σB−K IB−K σFe−L

INd−M σB−K IB−K σNd−M

=

=

= 0.1828

4.189 x 103 9.85×10−20 9.998 x 105 2.75×10−21 104

9.85×10−20

1.30 x 9.998 x 105 1.56×10−21

=0.15

=0.8225

We also know that NFe + NNd + NB = 1. We can then calculate the concentrations in the alloy as NFe = 7.6 at %, NNd = 41.7 at%, and NB = 50.7 at %.

9.4.4 Microdiffraction Even though microdiffraction involves elastic scattering, we discuss it here as it requires using a fine probe incident beam, such as that encountered in an AEM or a dedicated STEM. We motivate the use of microdiffraction, often encountered in the TEM as convergent beam electron diffraction (§8.6.3), using two simple arguments. First, as seen in §9.7.2.1 and Figure 9.2.9, a selected area aperture of diameter, ϕ∼ 10 μm, located in the first image plane of the OL, is used to select the area of the specimen for diffraction with an incident parallel beam. Then, the selected area on the specimen has a nominal diameter of ϕ/M, where M is the magnification of the OL, which is usually of the order of 20–50 times. This gives a minimum selectable area on the specimen of ∼0.2–0.5 μm. Many materials’ microstructures, such as grain boundary phases, nanoparticles, thin film heterostructures, semiconductor device components, etc., with feature sizes smaller than 0.2–0.5 μm require microdiffraction methods with focused beams for effective structural analysis. Unlike broad beam methods, e.g. selected area diffraction, that provide information averaged over larger volumes, microdiffraction methods provide local structural information defined largely by the size of the probe. Second, the spherical aberration of the OL also contributes to the size of the actual specimen area included in the diffraction. Even though the selected area diffraction aperture is conjugate to the specimen, there is a discrepancy or displacement, d, between the area selected in the image plane and where exactly in the specimen the diffracted beams that pass through the aperture originate from. To a good first approximation (Hirsch et al., 1977), d is given by d = CS θ 3 + (f ) θ

(9.4.32)

OUP CORRECTED – FINAL

Analytical Electron Microscopy (AEM) and Related Spectroscopies 649 where CS (∼1–6 mm) is the spherical aberration coefficient of the OL, θ is the scattering angle of the electron beam, and f is the defocus of the lens. Thus, d can be significant for large Bragg angles, θ, and the associated diffracted beams may not originate from the same area as that selected by the aperture. Note that f may be positive or negative and since CS θ 3 is always positive, the overall error in selecting the area for diffraction can be minimized by choosing f carefully; however, this approach is tedious and not commonly used. Example 9.4.4 You are imaging an orthorhombic crystal (a = 8 Å, b = 7 Å, c = 6 Å) using a 200 keV TEM with CS = 1.2 mm. What is the displacement between the area selected and the exact area from which a (111) diffracted beam arises, if the defocus is – 4000 Å? Solution: From Table 4.1.2, with (hkl) = (111), d 111 = 3.9587 Å. From (9.2.3b), λ (200 keV) = 0.025 Å. Thus θ = 0.1809. Hence, from (9.4.32), we get d = 7.03 μm. This displacement is quite large!

For these two reasons, microdiffraction methods, primarily CBED (introduced in §8.6.3) using a focused probe ∼1–10 nm in diameter, and a convergence semiangle, α, of the order of the Bragg angle, i.e. α = θ B , are preferred for structural characterization of microstructural features in the sub-μm length scale. Other salient applications of CBED in materials characterization include the following. 1. As we have seen in X-ray and EELS microanalysis, absolute quantification of concentration, (9.4.26), requires determining the local specimen thickness, t. As discussed earlier (§8.6.3; Fig. 8.6.11), using the two-beam CBED method, the fringes seen in the discs for crystalline materials can be used to obtain the local thickness. This same convergent probe used for microanalysis, without any change in the TEM/STEM operating conditions, can be used for the thickness measurements. 2. The radii of the higher-order Laue zones (HOLZ; Fig. 8.6.10c) can be simply related to the crystal periodicity along the beam direction. For a HOLZ diameter, Gn , in units of Å–1 , measured from the CBED pattern, where n is the order of the ring, it can be shown that  Gn =

2

2π nH − n2 H 2 λ

(9.4.33)

where H = 2π/duvw is the spacing of the reciprocal lattice along the electron beam direction. Neglecting the second order term, the period, duvw , of the crystal in the beam direction is then given by

OUP CORRECTED – FINAL

650 Transmission and Analytical Electron Microscopy duvw =

8π 2 G1 λ

(9.4.34)

where G 1 is the diameter of the first-order Laue zone (FOLZ) ring. Note that the unit cell parameters, including angles (Fig. 8.6.3) normal to the beam direction can be obtained from the analysis of the ZOLZ pattern. 3. Under favorable conditions the space group of the crystals can be determined from CBED patterns [80]. Such analysis involves details of the contrast within the discs of the CBED pattern, which requires somewhat thicker crystals. However, for thin crystals, an elegant method using a STEM unit and rocking the electron beam produces large-angle CBED patterns without overlapping of neighboring discs [81, 82]. In addition to §8, further details of electron microdiffraction, including CBED, can be found in specialized texts on TEM, including Williams and Carter (1996), Joy, Romig, and Goldstein (1986), and Reimer (1993).

9.5 Select Applications of TEM 9.5.1 Electron Tomography A TEM image is essentially a two-dimensional projection of the thin foil specimen along the direction of the primary beam. Thus, features within the depth of the specimen cannot be separately resolved. Stereo pairs can be recorded to resolve them if they are discrete features, but if they are continuously varying functions of mass or density they cannot be easily resolved. One way to overcome these limitations and obtain three-dimensional information with true depth sensitivity is by electron tomography. By tomography we mean the reconstruction of a threedimensional image of the object based on a series of two-dimensional projected images of the specimen acquired in different directions. The processing or reconstruction of the three-dimensional image from a set of two-dimensional projected images, critical for the successful implementation of tomography, involves two principal tasks: 1) alignment of the two-dimensional projected images with respect to each other, such that they can be referred to a common three-dimensional coordinate system describing the object to be reconstructed, and 2) computation of the three-dimensional image based on a reconstruction algorithm using one of the many approaches available. The technique of X-ray computed axial tomography (CAT), now common in diagnostic medicine, firmly established the principles of tomography as it is

OUP CORRECTED – FINAL

Select Applications of TEM 651 used today [83]. In this method, a thin fan-like beam is used to illuminate a slice of the object; the source and a linear detector placed behind the object are moved together to obtain a series of one-dimensional images/projections, which are then reconstructed to form a two-dimensional image. Finally, to obtain a threedimensional image, such two-dimensional image sets along different directions are recorded and reconstructed. Electron tomography in a TEM considers each image to be included in the reconstruction as a projection of the object along the primary beam direction. This important assumption is justified by the large depth of focus (9.2.8), which is so much larger than the diffraction-limited image resolution (9.2.3a). Moreover, since the specimen is finite, the full structural image at a predetermined resolution can be obtained by recording a series of projected images by sequentially tilting the specimen about a single axis over the largest range of angles possible (preferably 180◦ ) and maximizing the number of images acquired in the series (Fig. 9.5.1a). The relationship between the resolution, δ, and the increments in the tilt angle, α, are related to the diameter, D, of the object, and the number, N, of equally spaced projections included in the reconstruction is given by [85]: δ = Dα = π D/N

(9.5.1)

Thus, for the reconstruction of an object 25 nm in diameter (a typical nanoparticle, say), with a resolution of 0.25 nm (= 2.5 Å) will require 100 projections and a tilt increment of 1.8◦ . However, the TEM specimen geometry (extended in xand y-, but limited to the thickness in the z-direction) can be better sampled by nonequal tilt angle increments [86]. In practice, the difficulty in tilting the specimen over the full angular range causes distortions in the reconstructed image due to the missing data (Fig. 9.5.2) [87]. Second, in acquiring the large number (∼100) of projected images required for reconstruction, the specimen is subject to a large electron dose. To minimize the dose and exposure time, automated acquisition routines, especially with high precision, computer-controlled goniometer specimen stages are used [88–90]. Finally, as mentioned, each individually projected image has to be accurately aligned, i.e. tilted to a common tilt axis by changing the spatial position (x-y) and the rotational (θ) orientation of the specimen. To do so, a series of sequential crosscorrelations [91], and least-squares tracking of fiducial markers [92] are employed. For reconstruction (Fig. 9.5.1b), a real space back-projection method [93] is widely used. This involves projecting each two-dimensional image along the original tilt angle, and the superposition of these back-projected images gives the reconstructed image. However, the single tilt-axis geometry favors the sampling of low-frequency components of the image over the high frequency ones. This error is overcome by using a “weighting filter”, an approach that is referred to as the weighted back-projection method (Fig. 9.5.1b). Alternatively, iterative reconstruction methods [94, 95]—based on the principle that a reprojection of the threedimensional reconstructed object along the original projection angle/direction

OUP CORRECTED – FINAL

652 Transmission and Analytical Electron Microscopy (a) Data acquisition: projection of 2D images along the beam direction

(b) Tomographic reconstruction using consecutive 2D slices Raw backprojection Single slice

Apply weighting filter

Apply iterative reconstruction

or

R-weighted filter

SIRT with 10 iterations

(c) 3D visualization and segmentation

By slice

Slice in y-z

Slice in x-y

Slice in x-z

By iso-surface

By voxel projection (max-intensity)

Figure 9.5.1 The method of electron tomography. (a) The object is sampled by obtaining a tilt series of projected images, which are then aligned to a common tilt axis. (b) The three-dimensional image is reconstructed from the two-dimensional projected images using a back-projection algorithm. The raw back-projection shows significant blurring, which is overcome by appropriate frequency weighting or using an iterative approach to the reconstruction. (c) The reconstructed three-dimensional image of the object is visualized in a number of different ways including slicing, surface rendering, or voxel projection. Adapted from [84].

OUP CORRECTED – FINAL

Select Applications of TEM 653 5° increments

±90°

± 80°

± 70°

± 60°

± 50°

Original Image

2° increments

should be identical to the original image—are used, particularly for noisy and under-sampled images [96]. The key projection requirement for tomographic reconstruction is that the intensity in the image show a monotonic relationship with thickness of the specimen. For biological [90] and amorphous specimens, mass-thickness contrast (§9.3.2) satisfies this requirement. For crystalline materials, especially high-Z, such as heterogeneous catalyst particles, tomographic reconstruction has been successfully demonstrated using a set of tilt-series images obtained with a high angle annular dark-field (HAADF) detector (§9.3.4) in a dedicated STEM. The intensity in such Z-contrast images is linearly dependent on depth [97], and as mentioned (9.3.11), the differential cross-section is proportional to Z 2 . Electron tomography with chemical sensitivity has also been demonstrated using projected elemental maps, obtained with an EFTEM (§9.4.2.6), along different direction. Figure 9.5.3 illustrates EFTEM tomography of nanoscale precipitates in intermetallic alloys. Electron tomography is widely used in biology with particular emphasis on the reconstruction of images of single particles of complex macromolecules [90]. In fact, electron tomography, as a means to construct three-dimensional images from two-dimensional projections, owes a significant part of its development to the biological community. This is because electron tomography bridges a critical resolution gap between methods providing atomic resolution images and other microscopic methods, such as confocal light microscopy (§6.8.5) that allow imaging of living cells. Even though preparation of biological specimens involves embedding the cellular structures in vitrified ice, the resolution of ∼2–5 nm achievable in electron tomography, allows the possibility of imaging individual macromolecules to obtain information about their spatial relationship inside the cells.

Figure 9.5.2 Illustration of the effect of tilt range (missing data shown as a wedge in the middle row) and tilt increments (top: 5◦ increments, and bottom: 2◦ increments) on the reconstructed image. The reconstructed image is almost identical to the original when the tilt range is ±90◦ and tilt increment is 2◦ ; however, when tilt range is ±50◦ and tilt increment is 5◦ , there is no similarity. Adapted from [90].

OUP CORRECTED – FINAL

654 Transmission and Analytical Electron Microscopy (a)

(b)

2D BF TEM

(c)

3D Cr map

2D Cr map grain boundary

400 nm

180 nm Figure 9.5.3 Electron tomography of precipitates in intermetallic alloys using EFTEM images. (a) A series of surface rendered images following reconstruction of yttria precipitates in NiAl. (b) Images of grain-boundary carbide precipitates in 316 stainless steel showing a BF image (top) and a Cr-EFTEM map (bottom). Diffraction effects obscure the shapes of the carbide precipitates in the BF image, whereas the Cr-EFTEM map shows unambiguous contrast only from the precipitates; the latter are more suitable for 3D tomographic reconstruction. (c) The reconstructed tomographic image (voxel projection) showing the precise shape and orientation of the precipitates with respect to the matrix. Adapted from [84].

The total mean free path, tot , combining elastic and inelastic scattering in vitrified ice is ∼100 nm at E 0 = 100 keV. Typically, biological specimens (TEM sections) are several 100 nms in thickness, and so multiple scattering is always observed. Further, biological specimens (cells) and vitrified ice are largely composed of light elements (Z < 10), and hence, inelastic scattering in the forward direction dominates over elastic scattering (§9.4.1). This leads to a strong inelastic background in the images, and the associated noise causes significant blurring and obscures finer details in the images of macromolecules. As a result, EFTEM have been extensively used in biological imaging to filter out the majority of the inelastic scattering and using only the zero-loss peak to improve contrast in the images. However, the associated large doses required for EFTEM mapping limits its use in tomographic imaging. Many strategies including use of lower acceleration voltages to reduce knock-on damage (§5.3.2.2), the use of a FEG source with a narrow primary energy spread, and a liquid-He specimen stage to reduce the radiation damage to the specimen, are currently being employed [90]. In spite of these dose concerns, by combining EFTEM with electron tomography, three-dimensional distribution of elements, such as P, were obtained in thin sections of the nematode Caenorhabdites elegans, prepared by high-pressure freezing and plastic embedding [98]. The sections were sufficiently thin to allow jump-ratio imaging (see §9.4.2.6 and Fig. 9.4.13) using energy losses above and below the P L 2,3 edge. Subsequently, EFTEM projection maps (P jump ratio

OUP CORRECTED – FINAL

Select Applications of TEM 655

(a)

(b)

Figure 9.5.4 Topographic reconstruction of phosphoros distribution in a section of the C. elegans cell. (a) Volume rendered image of rows of ribosomes along stacks of endoplasmic reticulum membranes. (b) Higher magnification image of the phosphoros distribution shows individual ribosomes located at different heights within the section. Scale bar = 20 nm in both images. Adapted from [98].

images) were taken at 5◦ intervals over an angular tilt range of ±55◦ for two orthogonal tilt directions. Colloidal gold particles were used as fiduciary markers to align the images. A reconstructed three-dimensional image of the phosphoros distribution (Fig. 9.5.4) in the surface rendering, clearly shows features 15–20 nm in diameter that were identified as ribosomes distributed along the stacked membranes of the endoplasmic reticulum and in the cytoplasm. Surprisingly, these specimens were able to withstand a high electron dose of ∼107 e/nm2 in the process of image alignment, acquisition, and tomographic reconstruction. Complementing this, correlative imaging methods using dual function probes, such as FluorogoldTM ,14 hold much promise to identify specific bound elements and biological molecules in cellular structures [77]. For example, Fluorogold labels can detect specific proteins in the fluorescence optical microscope, followed by the localization of the same proteins in a TEM using the electron-dense gold labels [99], combined with electron tomography to image them in three dimensions.

9.5.2 Analysis of Defects: Dislocations and Stacking Faults One of the most well-developed applications of TEM is to investigate the diffraction contrast associated with images of dislocations (§4.1.9.2), stacking faults (§4.1.9.3), and other defects such as grain boundaries, interfaces, etc. In defective crystals the positions of atoms are slightly displaced (by the vector, R) from their original sites (defined by r) in the ideal crystal (Fig. 9.5.5); thus, their positions are given by (r + R). Then, following (8.5.10), for the atoms in a crystal that are misoriented by s, from the exact Bragg condition in diffraction,

14 Hydroxystilbamidine (trade name FluoroGold) is a fluorescent dye that emits different frequencies of light when bound to DNA and RNA.

OUP CORRECTED – FINAL

656 Transmission and Analytical Electron Microscopy (a)

(b)

(c)

R = constant

Dislocation Perfect Crystal “column”

R(z)

Stacking Fault

Figure 9.5.5 The effect of defects on a columnar beam in (a) a perfect crystal, and a crystal with (b) a dislocation with the lattice displacement varying along z, and (c) a stacking fault with a constant lattice displacement in the lower half of the thin foil. Adapted from Fuchs, Oppolzer, and Rehme (1990).

and are defective with an additional displacement, R, the phase angle, φ g , for any reflection, g, is given by φg = (g + s) · (r + R) = [(g·r) + (g·R) + (s·r) + (s·R)]

(9.5.2)

The term (s·R) is a product of two small vectors and can be neglected. The term (g·R) is an integer and does not affect the phase change. The two remaining terms, (g·R) + (s·R), added together contribute to the total phase shift for either any lattice defect, or misorientation from the Bragg angle, θ B , respectively. Further, at the exact Bragg condition, s = 0, the scalar product, (g·R), dominates the phase angle. Thus, for specific Bragg reflections, g, where the contrast of the lattice defect satisfies the criterion g·R = 0

(9.5.3)

its image disappears, and under these conditions, both the direction and magnitude of R can be determined. For stacking faults (Fig. 8.5.8c), |R| = constant; then, when the condition (g·R) = 0 is satisfied, the stacking faults become invisible. In the case of dislocations (Fig. 8.5.8b), R is a function of position, r, and specifically for thin foils, it is a function of depth, R(z). Further, for a pure screw dislocation, by definition the Burgers vector, b, is parallel to the displacements,

OUP CORRECTED – FINAL

Select Applications of TEM 657 R, for all atoms (Fig. 4.1.20b), which are all along the dislocation line defined by ˆ Then, R can be replaced by b, and when the condition g·b the unit vector, u. = 0 is satisfied, pure screw dislocations become invisible. In the general case, the displacement, R, in an isotropic solid for a general (mixed) dislocation, where b ˆ is given by is not parallel to u, 1 R= 2π





  1 bα + be + b ⊗ uˆ [2 (1 − 2ν) ln |r| + cos 2α] 4 (1 − ν)

(9.5.4)

where be is the edge component of the Burgers vector, ν is the Poisson ratio, and polar coordinates (α) are used. Then, for a pure edge dislocation b = be , and (g·R)   involves two terms, i.e. (g·be ) and g·be ⊗ uˆ . The second cross-product term arises from the buckling of the glide plane by the presence of the edge dislocation, and can complicate the analysis of dislocations with an edge component, as they may not go completely out of contrast. Nevertheless, the (g·b) = 0 criterion can be generalized to all dislocations and stated as a rule; i.e. if the Burgers vector, b, of the dislocation is perpendicular to the active diffraction vector, g, no diffraction contrast will be observed from the dislocation and it will be invisible. This is illustrated schematically in Figure 9.5.6. Further if two Bragg reflections g1 and g2 , satisfy g·b = 0, then g1 ⊗ g2 b. In practice, the dislocations become invisible even when g·b < 1/3. In addition, the analysis and determination of b is more complicated because the dislocations may not be invisible, even when g·b = 0, if there is a contribution of the term g·be ⊗ uˆ = 0. Further details of such analysis can be found in Edington (1976). Table 9.5.1 shows the values of various g·b combinations for perfect dislocations in FCC crystal. Orientations such as [110] are particularly useful because it gives access to g vectors of the form (002), (111), (220), (113), etc. Figure 9.5.7 shows

Bright Field

Dark Field or: 2-Beam Bright Field

Dark Field or: 2-Beam Bright Field

Figure 9.5.6 Schematic illustration of the diffraction contrast from dislocations, with b and lines in the plane of the image. Adapted from Fultz and Howe (2013).

OUP CORRECTED – FINAL

658 Transmission and Analytical Electron Microscopy Table 9.5.1 Values of g·b for perfect dislocations in FCC crystals. Plane of dislocation     111 or 111     111 or 111     111 or 111   (111) or 111   (111) or 111   (111) or 111

b 1 2 1 2 1 2 1 2 1 2 1 2

g

111

111

111

002

020

220

[110] [101] [011]   110   101   011

0 1 0 1 0 1

0 0 1 –1 –1 0

1 0 0 0 1 –1

0 1 1 0 –1 1

–1 0 –1 1 0 1

0 1 –1 2 1 1

Adapted from Thomas and Goringe (1980).

– 002

(a)

(b)

– – 111

(c)

– 113

Figure 9.5.7 Bright field images in the two-beam condition of edge dislocations in a TiAl alloy. The corresponding g vectors are shown. 0.1 μm

Adapted from Fultz and Howe (2013).

g·b analysis of edge dislocations in TiAl. It can be seen that the dislocations are primarily edge type with b = 12 (110). Sharper images of dislocations, at better resolution (Fig. 9.5.8) can be obtained by weak-beam dark-field imaging using the g-3g diffraction condition (Fig. 9.3.5b). Further details of weak beam imaging can be found in [100]. Diffraction contrast of stacking faults is not discussed here, but similar analysis including establishing whether the stacking faults are intrinsic or extrinsic, as well as practical details and pitfalls, is found in Edington (1976), Williams and Carter (1996), and Fultz and Howe (2013).

9.5.3

Thin Films and Multilayers: An Example

We now revisit the TEM analysis of the thin film (Mo5.4nm Si1.3nm )n multilayer, introduced in §8.7.5, and answer some of the question raised earlier to provide a complete picture of its structure, composition, and microstructure. The BF image (Fig. 9.5.9a) of a cross-section specimen of the multilayer clearly shows a substrate

OUP CORRECTED – FINAL

Select Applications of TEM 659

Figure 9.5.8 Dislocations in silicon imaged with a (a) strong 220 diffraction beam (inset), and (b) weak-beam 220 dark field, g-3g, condition. Notice the significantly better resolution in the weak-beam image. (a)

(b)

100nm

(a)

BF

(b) Counts

80

EDXS-line scan

Si

Cr

Mo

20

40

60 Position (nm)

EFTEM Si-K Map

(e)

20 0 0.00

(d)

Si Mo Cr

60 40

Adapted from [100], which also describes weak-beam imaging in detail.

100nm

80

(c)

100

SAD

HREM 5.7 nm 1.3 nm

Mo 011

Mo / a-Si multilayers, dMo = 5.4 ± 0.05nm, da-Si = 1.3 ± 0.05nm.

on which we can observe a thick first layer, followed by two layers that together repeat with a periodicity of ∼7 nm. Moreover, the thicker of the two repeating layers shows diffraction contrast within the layer, indicative of columnar growth, but the thinner layer shows no contrast, suggesting that it may be amorphous. Further, the selected area diffraction pattern (Fig. 9.5.9c) reveals additional specifics about the crystallographic structure of the layers. First, it should be noted that the diffraction pattern is comprised of two parts: reflections from the crystalline Mo layer, and superlattice reflections, shown in the magnified part of the central portion of the diffraction pattern with very fine spacing that correspond to the overall multilayer period. The Mo reflection can be indexed as (011), and shows some texture (formation of arcs) of the columnar grains; moreover, the Mo(011) reflection serves as an internal calibration. The fine superlattice spacing are 1/37.8 of the Mo(011) reflection, and hence the superlattice period is d 011 (∼ 0.1767 nm) × 37.8

Figure 9.5.9 Comprehensive analysis of a nominal [Mo5.4nm Si1.3nm ]n multilayer cross-section specimen using a 200 kV analytical electron microscope with Schottky emission gun, EDXS detector, and a post-column image filter. (a) Bright field image, (b) X-ray line scan, (c) selected area diffraction pattern, (d) energy-filtered image using the Si K edge, and (e) a high-resolution phase contrast image shows that the Mo layer in reality is slightly thicker (5.7 nm) than expected.

OUP CORRECTED – FINAL

660 Transmission and Analytical Electron Microscopy = 6.68 nm. Finally, the HREM phase contrast image (Fig. 9.5.9e) confirms the overall periodicity of the multilayer (which is to be expected), the high crystalline quality of the Mo layer as well as the amorphous nature of the Si layer, and most importantly that the Mo/a–Si interface is not sharp but atomically intermixed. What are the multilayers made of? To answer this question, we can turn to energy dispersive X-ray microanalysis using a 0.5 nm probe. This is possible because the microscope (Fig. 9.1.2) is equipped with a high-current density Schottky emission gun and a windowless energy dispersive X-ray detector. X-ray spectra were obtained by positioning the probe at different points along a straight line starting at the substrate and through the multilayer stack, and the characteristic X-ray intensities were plotted as a function of position (Fig. 9.5.9b). The analysis clearly revealed alternating layers to be Mo (5–5.5 nm) and Si (1– 1.5 nm); in addition, it also revealed an unknown Cr underlayer used in the growth that was not known before the TEM analysis. Finally, an EFTEM image (Fig. 9.5.9d) using Si K-edge (E ∼ 1800 eV) confirms the silicon distribution. It also confirms that EFTEM can obtain images with resolution of the order of 1 nm, even at such high energy-losses as the Si K-edge ( ∼ 1740 eV loss).

9.5.4 TEM in Semiconductor Manufacturing: Metrology, Process Development, and Failure Analysis Continued technological scaling of metal oxide semiconductor field effect transistors (MOSFETs), ubiquitous in integrated circuits (ICs), presents significant challenges in (a) determining precisely the size and extent of the ultrathin layers and interfaces that make up the individual transistor (metrology), (b) development of alternative high-k dielectrics to replace silicon dioxide as the gate oxide used to isolate the gate electrodes from the source-drain channel (process development), and (c) identifying the root cause of a failure site indicated by electrical measurements ( failure analysis) [101]. The narrowest feature in an IC is the gate oxide, or the thin dielectric layer that forms the basis of the field-effect device structure. Silicon dioxide is widely used and the physical thickness of this gate is of the order of 1 nm; however, alternative gate oxides, such as HfO2 , ZrO2 , etc., which can be thicker than 1 nm because of their higher dielectric constants, are in development. Nevertheless, in manufacturing of ICs, accurate measurements of the thickness of the gate oxide layer are necessary as even a 10 % decrease in its thickness will change the leakage current by one order of magnitude [101]. Moreover, a thickness of 1 nm of the oxide layer corresponds to about five silicon atoms spanning the layer [102]; of these five, at least two of the atoms will be at the silicon–oxide interface with electrical properties expected to be different from those in the bulk oxide. It is easy to recognize that such rapidly shrinking device features below the nanometer scale will push the required metrology well beyond the resolution limits of even the best SEMs, which is of the order of 5–7 nm (§10.2.4). Further, in manufacturing, process anneals at ∼1,000◦ C

OUP CORRECTED – FINAL

Select Applications of TEM 661 are involved; hence, it is important to study not only the stability but also the interactions between the dielectric layer and the silicon channel of the device at elevated temperatures. Generally, most electrical failures of the device to be resolved require the physical and chemical characterization of defects buried deep inside the structure. SEM only provides image contrast based on surface topography and/or difference in the atomic weights of the elements distributed on the surface. In practice, a buried defect will not change the surface topography, and by extension the contrast in imaging with an SEM, unless a decorative etch is used; this is also a destructive process and hence is not applicable. The last resort to address these three issues in semiconductor manufacturing is a TEM, which has the required resolution and is also a versatile instrument providing structural, crystallographic, and chemical information (Fig. 9.1.2). However, the most important key to addressing these problems with a TEM is the development of site-specific specimen preparation of the device/wafer using a focused ion beam instrument [103]. Specifically, the advanced dual column FIB-SEM instruments (Fig. 10.7.3) with separate high-resolution drift-free ion beam and electronbeam columns allow for the simultaneous imaging and sectioning of the specific defective portion of the IC device into an electron transparent plan-view or crosssection TEM specimen (Fig. 10.7.4c). Methods of TEM specimen preparation are discussed briefly in Section 9.6), and we now present some representative examples of MOSFET structures studied by TEM [101]. Once the proper TEM specimen is prepared, conventional BF imaging (Fig. 9.5.10) can be used to evaluate the profile and critical dimensions of the components of the MOSFET, including the poly-Si gate, nitride spacer, active Si trench depth, etc. Such measurements of the critical dimensions are often used to tailor the etch processes to improve the gate profile, including tapering and footing, and ensure that the device meets the performance specifications. Again, if a damage-free ultra-thin (thickness E. Adapted from Weisendanger (1994).

Impenetrable Barrier

Eb

(b) Quantum Mechanics Tunneling Effect

Eb E

OUP CORRECTED – FINAL

Introduction 747 tunneling can take place in a controlled and predictable manner across a vacuum gap, between a sharp metal tip/needle and a surface, provided the tip is scanned at a distance of less than ∼1 nm from the surface. This is because the tunneling current in an STM decays over a length scale of an atomic radius and flows from the terminating atom at the apex of the tip to a single atom on the surface (§11.3.1), or vice versa. Thus, atomic resolution is integral to the physics underlying the tunneling contrast mechanism in a STM. However, to achieve this resolution in practice, STM requires highly sensitive piezoelectric2 actuators (§11.6) to move the tip orthogonally in three directions with control over atomic dimensions. It also requires high mechanical stability (no vibrations) of the experimental setup. When both these criteria are satisfied, as in contemporary STM design and construction (Fig. 11.3.1), an electronic controller can be used to adjust the tip–surface separation such that either a constant/preset tunneling current with variable separation/height, or a constant height with variable tunneling current, is maintained throughout the scan. Then the separation or current, respectively, recorded as a function of lateral position, can be displayed at atomic resolution as a microscopic image of the surface. Strictly speaking, an STM maps a constant local density of states (§3.2) at the surface, which may be different from the surface topography. Moreover, it can also study the electronic structure of the surface by stopping the probe at any specific location and studying the tunneling current as a function of tip–surface voltage; this mode is known as scanning tunneling spectroscopy (STS, §11.3.2). Finally, the STM can also be used to manipulate or move atoms on surfaces as discussed in §11.3.3. All scanning probe microscopes are variants of the STM described (Fig. 11.1.2). In all cases, a local probe/tip senses a short-range interaction, be it a current or force, with the surface. The probe is scanned using piezoelectric scanners and the strength of the local interaction—the tunneling current, or a force—is recorded as a function of lateral position, and plotted as an image. The most important variant of the SPM is the scanning force microscope (SFM), commonly known as the atomic force microscope (AFM), which has the distinct advantage, unlike the STM, of being able to image nonconducting surfaces [2]. An SFM, as the name implies, measures the force between the tip and the surface by mounting the tip on a microfabricated cantilever that behaves as a force sensor. Either the deflection of the cantilever or the change in its dynamic response, as a result of the local tip–surface forces, is recorded to form the image. SFMs take advantage of various possible tip–surface interactions or forces, with specific designs optimized for sensing each one (Fig. 11.1.2). In the remaining sections of this chapter, we present a comprehensive introduction to the instrumentation, physical principles, and applications of SPM. Our emphasis mainly is on STM and scanning/AFM. Further details on SPMs can be found in numerous, well-written textbooks and monographs, including Chen (1993), Wiesendanger (1994), Meyer, Hug, and Bennewitz (2004), and Eaton and West (2010).

2 Piezoelectric materials show an electric polarization on the application of stress. This linear coupling between electrical and mechanical energies is described as a tensor, Pi = dijk σjk , where Pi is the induced polarization, and σ is the applied stress. Strictly speaking, here we are interested in the inverse piezoelectric effect, xij = dkij Ek , where the same coefficients, dijk , are used to relate the applied electric field, E, to the induced strain, x.

OUP CORRECTED – FINAL

748 Scanning Probe Microscopy SNOM

SNAM

SICM

PEMSA

hf

e hf

STMiP

LSTM

It

ECSTM

It

It

T

hf

hf

SNM 2 2 T1

SThM

SPotM

SCM

SSRM

I C

R

T2 Figure 11.1.2 Various designs of scanning probe microscopes; all these methods are characterized by a local probe, registering a local signal/interaction, as it is scanned over the surface to form an image. Top (l to r): Scanning near-field optical microscope (SNOM), Scanning near-field acoustic microscope (SNAM), Scanning ion conductance microscope (SICM), and Photoemission microscope with scanning aperture (PEMSA). Middle (l to r): STM with inverse photoemission (STMiP), Laser scanning tunneling microscope (LSTM), Electrochemical STM (ECSTM), and Scanning thermal microscope (SThM). Bottom (l to r): Scanning noise microscope (SNM), Scanning tunneling potentiometry (SPotM), Scanning capacitance microscope (SCM), and Scanning spreading resistance microscope (SSRM). One of these (SThM) is discussed further in this chapter (§11.8.3), but detailed discussions of the rest can be found in Meyer, Hug, and Bennewitz (2004), from which this figure is adapted.

OUP CORRECTED – FINAL

Physics of Scanning Tunneling Microscopy (STM) 749

11.2 Physics of Scanning Tunneling Microscopy (STM) 11.2.1 Elastic Tunneling Through a One-Dimensional Barrier The basic physical principles of operation of an STM can be understood from the simple model of elastic tunneling through a one-dimensional potential barrier of width, s (Fig. 11.2.1). By simply matching the wave function (see Wiesendanger (1994) for a detailed derivation) on either side of the barrier, it can be shown that the transmission coefficient, T, defined as the ratio, T = jt /ji , where jt is the transmitted (region 3) and ji (= k/m) is the incident (region 1) current density, is given by 16k2 χ 2 −2χ s 2 e k2 + χ 2

T= 

where k =

√

(11.2.1)

2me E/, and the decay rate, χ , is given by √ χ=

2me (E0 − E) 

(11.2.2)

Here, a strongly attenuating barrier, with χs >> 1, is assumed. It is now easy to see that the transmission coefficient, T, is dominated by the factor exp(–2χs). As such, T is dependent on the product of the barrier width, s (Å), and the effective barrier height, (E0 -E) in eV, but not on the exact shape of the barrier. Note that if we make the reasonable assumption that the effective barrier height is of the order of the work function, , and use Au (∼5 eV) as an example, we can see that a change in barrier width by 1 Å leads to one order of magnitude change in the barrier transmission (see Problem 1). This very high sensitivity of the tunneling current to the barrier width is the underlying motivation for STM. Example 11.2.1: A one-dimensional square potential barrier has a height of 10 eV and a width of 3 nm, and electrons with kinetic energy of 7 eV are trying to tunnel through this barrier. (a) What would be the decay rate of these electrons? (b) Would this be a strongly attenuating barrier? (c) What is the transmission coefficient for the tunneling of these electrons? (d) What will be the new transmission coefficient for the same electrons if the barrier width is reduced to 1 nm?

1

k

2

3

E0 E

k

0

s

Figure 11.2.1 A one-dimensional potential barrier of height, E0 , and width, s. An electron of energy, E, and wave vector, k, impinging on the barrier in region 1, has a finite probability of tunneling elastically to region 3.

OUP CORRECTED – FINAL

750 Scanning Probe Microscopy

Solution: (a) The decay rate (11.2.2) is  1/2     χ = 2 × 9.1 × 10−31 × (10 − 7) 1.6 × 10−19 1.05 × 10−34 = 8.90 × 109 m−1 . (b) The attenuating barrier for a barrier width, s = 3 nm, is χ s = 8.9 × 109 × 3 × 10−9 = 26.7  1. Hence, we can assume that this is a strongly attenuating barrier. (c) The transmission coefficient, (11.2.1), can be written as T = T0 exp (−2χs), where by rearranging terms, it can be shown (the derivation is left as an exercise to the reader) that T0 = 16E (E0 − E) /E0 2 = 16 × 7 × 3/102 = 3.36. Thus, for a barrier width of 3nm, T = 3.36 exp (−2 × 26.7) = 2.16 × 10−23 , which is a very small number, indeed. (d) If the barrier width is 1 nm, the new transmission coefficient is   T = 3.36 exp −2 × 8.9 × 109 × 1 × 10−9 = 6.25 × 10−8 . In other words, reducing the barrier height by a factor of three, or a small change in the barrier height, increases the transmission coefficient by 1015 , i.e. leads to a very large change in the transmission coefficient.

11.2.2 Quantum Mechanical Tunneling Model of the STM In STM, the tunneling barrier width, s, is given by the vacuum gap, z, between the tip and the specimen surface, and the work function, , replaces the barrier height, E 0 . However, in addition to the effective barrier height, ( – E), and the tip–surface separation, z, the tunneling current, IT , also depends on the applied bias voltage, Vb , and the local density of states, Ns (E F ), of the specimen at the Fermi level, and is given by √

√ 2me ( − E) z ∝ Vb Ns (EF ) e−1.025 z IT ∝ Vb Ns (EF ) exp −2 

(11.2.3)

This can be best understood by considering the energy level diagram of the tip and the specimen, starting with them being far apart to be electronically

OUP CORRECTED – FINAL

Physics of Scanning Tunneling Microscopy (STM) 751 Tip EV

T

Vacuum

(a)

EFT (b)

Specimen S

T

EFS

N (E)

EV EFT

EV

S

EV EFS

z (c)

EV T EFT

S

EV EFS +Vb

(d)

EV EFT

T

S

EV EFS –Vb

independent (Fig. 11.2.2a). When the tip is brought close to the specimen surface with a small vacuum gap, z, they are in equilibrium and their Fermi levels are aligned equal (Fig. 11.2.2b). Upon applying a bias voltage, Vb , to the specimen the energy levels will undergo a rigid shift in energy by the amount |eVb |. This can be downward (Fig. 11.2.2c) for positive polarity of the bias, or upward (Fig. 11.2.2d) for negative polarity of the bias. Thus, for positive specimen bias (Fig. 11.2.2c) the tunneling current flows from the tip to the specimen, and probes its local unoccupied density of states as a function of bias voltage. On the other hand, for negative bias (Fig. 11.2.2d), the electrons tunnel from the occupied states of the specimen to the unoccupied states of the tip. Hence, by varying the sign (polarity) and magnitude of the bias voltage, both the occupied and unoccupied local density of states of the specimen can be probed. This is called STS and is discussed further in §11.3.2.

Example 11.2.2: An STM tip is used to probe an Au surface with a bias voltage of 2V. If the tip specimen height changes from 1 Å to 2 Å, what will be the percentage change in the tunneling current? Solution: The work function for Au is  = 5 V and Vb = 2 V. Applying (11.2.3), we calculate the percentage change as

Figure 11.2.2 Energy level diagram for the tip (left) and the specimen (right) in a scanning tunneling microscope. (a) The tip and specimen, when far away are electronically independent. The work function, , the vacuum level, EV , and the Fermi level, EF , for both the tip and specimen, as well as the density of states, N(E), for the specimen are shown. (b) The tip and specimen are in equilibrium, i.e. EFT = EFS , and are separated by a small vacuum gap, z. (c) Positive specimen bias voltage, +Vb , with electrons tunneling from the tip to the specimen. (d) Negative specimen bias, –Vb , with electrons tunneling from the specimen to the tip. Adapted from Wiesendanger (1994).

OUP CORRECTED – FINAL

752 Scanning Probe Microscopy

IT2 − IT1 IT1

× 100 =

e−1.025

√

2

− e−1.025

e−1.025

√

√





 √ √  × 100 = 1 − e−1.025 2−  100

= (1 − 0.387) 100 = 61% which is truly substantial.

11.3 Basic Operation of the Scanning Tunneling Microscope Figure 11.3.1 shows a schematic diagram of the STM. It consists of a sharp metal tip that is brought in close proximity to a surface, creating an atomic-scale vacuum gap, using the xyz piezo drives. A bias voltage applied between the tip and the

xyz Piezo Scanner

z y

x

High-Voltage Amplifier

IT Probing Tip

Feedback

Specimen

Vibration Isolation Figure 11.3.1 Schematic representation of a scanning tunneling microscope. The xyz piezo-scanner moves the tip in a controlled manner over the specimen surface. The feedback loop is used to keep a constant tunneling current, and the height, z, is recorded as a function of the scanned coordinates, x and y, to give the image. Note that a high voltage amplifier to drive the piezo scanner and good vibration isolation are required for atomic-resolution imaging. The entire microscope portion, enclosed in the dashed line, is kept under UHV to avoid surface contamination. Adapted from Meyer, Hug, and Bennewitz (2004).

OUP CORRECTED – FINAL

Basic Operation of the Scanning Tunneling Microscope 753 specimen results in electrons tunneling through the vacuum gap. Typically, the tunneling current is in the nA to pA range. This current signal is amplified and used as the input voltage for the feedback loop to the piezo drives, which adjusts the tip height to maintain a constant tunneling current as the tip is scanned over the surface. The tip height (or, in reality, the feedback voltage to the piezo drive) is plotted as a function of position to form the image. This imaging method is called the constant current mode. Alternatively, the tip can be moved at constant height and the variations in the tunneling current measured to form the image. These two modes of STM imaging are discussed in §11.3.1. The tip can also be physically stopped at any position, (xi , yi ), and the bias voltage, Vb , or tip–surface separation, z, can be varied up or down to measure the variation of a local property such as the barrier height (work function). This spectroscopic mode is discussed in §11.3.2. Finally, an STM can also be used to manipulate atoms on a surface, by picking them up individually with the tip and moving them by translating the tip position; this is discussed in §11.3.3.

11.3.1 Imaging We assume, without any loss of generality, that the specimen is fixed on the stage and the STM tip is scanned in the x-y plane, even though there are some STMs that do it the other way around. Anyway, for the purpose of our discussion the two arrangements are equivalent, and we now discuss the two common modes of STM imaging. In the constant height mode (Fig. 11.3.2a), the tip does NOT move in the vertical (z) direction as it is scanned laterally (x-y) over the specimen surface. The tunneling current across the vacuum gap is recorded, which is inversely related to the local tip–surface distance, and varies as the surface profile of the specimen. However, the dependence of the tunneling current on the surface profile is not linear, and depends exponentially on the tip–surface separation as described by (11.2.3). In addition, the current is also dependent on the electronic structure, or the density of states, of the specimen (Fig. 11.2.2). From an experimental point of view, the constant height mode works well for flat specimens, where the roughness or corrugation is smaller than the average tip–surface distance. Further, two practical issues are encountered: there is always a vertical drift of the tip due to thermal effects and often the overall specimen plane is not parallel but inclined to the lateral scanning plane (x-y) of the tip. To work around these issues, the feedback system is turned on and operated at low sensitivity to keep the average current at a preset value. The low sensitivity ensures that the feedback mechanism does not interfere with the fine variations of the current caused by local changes in the surface profile; at the same time, it is fast enough to recognize the inclination of the specimen surface. In the constant current mode (Fig. 11.3.2b), the feedback system mentioned earlier is actively engaged and used to control the vertical position of the tip in such a way that a constant tunneling current (preset value) is always maintained.

OUP CORRECTED – FINAL

754 Scanning Probe Microscopy (a) IT 0

Figure 11.3.2 Two modes of STM imaging include (a) constant height mode, and (b) constant current mode. The tunneling current in both cases is shown in the inset (plot). The tunneling takes place between the atom at the apex of the tip and an atom on the surface of the specimen. Note that the topography measured is a convolution of the real surface with the physical dimension of the tip.

(b) IT 0 Tip

Specimen

Adapted from Colton et al. (1998).

3 Surface reconstruction and related nomenclature was introduced in §8.3. 4 An atom that lies on a crystal surface is called an adatom. It can be thought of as the opposite of a surface vacancy.

Then, the tip follows the surface profile assuming a constant energy barrier height. In practice, the voltage applied to the piezo drive in the z-direction controls the vertical tip position, and this voltage (plotted, as shown) is proportional to the surface profile. Note that in this mode, the scanning speed is limited by the reaction time of the feedback and scanning system. The constant height mode is used for atomically flat specimens, as the time constant of the feedback system is not significant in this mode of operation. However, if the surface roughness/corrugation is significantly larger than the tip– surface distance, there is a very high probability of the constant-height tip crashing into the specimen; in that case, the constant current mode is preferred. One of the very first demonstrations to highlight the capabilities of the STM was the study of the 7 × 7 reconstruction3 of the Si(111) surface, which was resolved by direct real-space imaging [3], using the constant current mode (Fig. 11.3.3). The original recorded image of two 7 × 7 unit cells is shown in relief in Figure 11.3.3a; the dark lines of minima highlighting the boundaries of the rhombohedral 7 × 7 unit cells are clearly seen. A top view of the relief is shown in Figure 11.3.3b, where the sixfold rotational symmetry is obvious. The crosses indicate the adatom positions4 that agree with an adatom model as shown in Figure 11.3.3c. In addition to the tunneling current, IT (x,y), its derivative, dIT /dz(x,y), often referred to as barrier height or work function signal, is recorded simultaneously. Taking the derivative of (11.2.3), it can be shown that √ dIT /dz ∝ −IT 

(11.3.1)

OUP CORRECTED – FINAL

Basic Operation of the Scanning Tunneling Microscope 755 (a)

(b)

(c)

Figure 11.3.3 STM constant current mode image of the 7 × 7 reconstruction of Si (111). (a) The image in relief, taken at 300◦ C, showing two unit-cells. (b) Top view of the relief, with the level of brightness indicating the altitude (but not to scale), and crosses representing adatom positions. (c) An adatom model of the silicon surface, with the grid indicating the 7 × 7 unit-cell. Adapted from [3].

A

D

Tunneling resistance (Ω)

10–6

B C 10–7

106

10–8

107

10–9 108

Tunneling current, IT (A)

E 105

1Å

Displacement, z, of the W-tip

Thus, the square of the normalized derivative signal, (dIT /IT ) /dz, is proportional to the height of the tunneling barrier, or to a first approximation, the work function, . However, in practice the values of [(dIT /IT ) /dz]2 that are measured are smaller than theoretical values because of surface contamination. Alternatively, measuring this derivative signal gives an indication of the cleanliness of the tip, and is often used as a check before imaging studies. Figure 11.3.4 shows one of the first such measurements of the barrier height distribution in a Pt specimen using a tungsten tip√[4]. Note that here the tunneling current (11.2.3) is fitted in the form ln IT = −A  z + B, and the slope is interpreted in terms of the work function.

11.3.2 Tunneling Spectroscopy We have seen (Fig. 11.2.2) how the local electronic density of states of the specimen contributes to the tunneling current. Note that the current increases significantly

Figure 11.3.4 The variation of the tunneling current as a function of the W-tip displacement in a Pt specimen. The derivative of the tunneling current gives a work function, ∼0.6–0.7 eV, for a normal surface (curves A, B, C). However, when the surface was repeatedly cleaved in situ in vacuum (curves D, E), a significantly steeper slope gives a value of ∼3.2 eV. Adapted from [4].

OUP CORRECTED – FINAL

756 Scanning Probe Microscopy if the bias voltage, Vb , allows electrons to tunnel to a maximum of the unoccupied density of states locally in the specimen. In addition, electrons at the highest energy—close to the Fermi level—experience the smallest effective tunneling barrier, as indicated by the magnitude of the arrows in Figure 11.2.2 c,d. This is also known as the barrier transmission coefficient, T(E,eVb ), which is a function of the energy, E, of the electron and the bias voltage, Vb . Thus, to first order, a typical tunneling experiment would yield a tunneling current that is a product of the tunneling coefficient, T (E, eVb ), and the local density of states, N(E), of the tip and specimen. Figure 11.3.5 is a schematic illustration of a spectroscopic tunneling experiment with the tip biased positively by +1 V, allowing the electrons to tunnel from the specimen to the tip. For the purpose of discussion, Figure 11.3.5a shows the densities of states by different curves for the specimen (slow, or low frequency wiggles) and tip (fast, or high frequency wiggles). Figure 11.3.5b illustrates the expected tunneling current for both positive and negative bias voltage. We see that

(a)

Tip

Adapted from Wiesendanger (1994).

Specimen

S

EF eVb T

EF

(b) Tunneling current

Figure 11.3.5 Schematic illustrations of the principles of tunneling spectroscopy using an STM. (a) The density of states (DOS) of the specimen (slow wiggles) and the tip (fast wiggles), as well as the applied positive bias voltage, Vb , leading to a tunneling current flow from the specimen to the tip, is shown. (b) The resultant tunneling spectrum, as a function of bias voltage, represents the DOS of the tip for negative bias, and the DOS of the specimen for positive bias. This signal is superimposed on a background (dashed line) arising from the transmission coefficient, T(E,eVb ), that increases monotonically, independent of the sign, with the magnitude of the bias voltage.

2.0 1.5 1.0 0.5 0 –1.0

–0.5

0

0.5

Specimen bias voltage, Vb (V)

1.0

OUP CORRECTED – FINAL

Basic Operation of the Scanning Tunneling Microscope 757 the transmission coefficient increases monotonically as a function of bias voltage and contributes a smooth background, on which the electronic structure (local density of states) is superimposed. We also see (Fig. 11.3.5b) that the resultant spectrum shows fidelity to the density of states of the specimen (slow wiggles) for positive bias voltage, and that of the tip (fast wiggles) for negative bias. In practice, the simplest method to obtain spectroscopic information in STM is to sequentially record topographical images at different bias voltages using the constant current method (§11.3.1). For analysis of the spectra, we assume that at any specific bias voltage only the electronic states at the Fermi level of the specimen and/or tip contribute to the tunneling current; in other words, the interference of the local geometry/topography with the electronic structure is neglected. More sophisticated experimental arrangements for spectroscopy, where the effect of topography can be removed, are discussed in Wiesendanger (1994) and Chen (1993). Again, one of the first applications of STS was the measurement of the bands of the Si(111) 2 × 1 reconstructed surface [5], using the differential conductivity, dI/dV, to directly measure the surface DOS. Data were obtained over the energy range −4eV – +4 eV, relative to the Fermi level, and showed the expected structure of two π-bonded surface bands and a new surface resonance 2.3 eV above the Fermi level. For those interested, further details can be found in [5].

11.3.3 Manipulation of Adsorbed Atoms on Clean Surfaces There are two common ways to manipulate adsorbed atoms on surfaces using an STM tip. This includes both vertical (normal to the surface) and lateral (along the surface) manipulation (Fig. 11.3.6).

(a)

(b)

Figure 11.3.6 Ways of manipulating adsorbed atoms on clean surfaces (a) Vertical manipulation followed by lateral motion of the tip. (b) Lateral manipulation that requires overcoming the resistance from the surface to the adatom motion, typically in the K – M range. Adapted from Meyer, Hug and Bennewitz (2004).

OUP CORRECTED – FINAL

758 Scanning Probe Microscopy

V V0 ψ3 ψ2 ψ1 x 0

L

Figure 11.3.7 A one dimensional potential well of height, V0 , and length, L. The first three allowed wave functions are also shown.

In the case of vertical manipulation, the tip is brought into contact or near contact with the adsorbed atom. The atom is then transferred from the surface to the tip by the application of a voltage pulse of appropriate magnitude and sign. For example, in the case of Xenon atoms, the adsorbed atoms move in the same direction as the tunneling current by a process called heat assisted electromigration [6]. If necessary, once the atom is transferred to the tip, the latter can be retracted from the surface and moved laterally by changing the tip position. Then, a reverse voltage pulse can be applied to release the adsorbed atom to the surface as illustrated in path (a) of Figure 11.3.6. Alternatively, by adjusting its vertical position the tip can be set to form a weak bond with the atom and then moved laterally, path (b), to “push” the adsorbed atom from one location to another. The threshold resistance to move the atom depends on both the atomic species and the nature of the specific surface. Typical values are 5 M for Xe on Ni(110), 200 K for CO on Pt(111), and 20 K for Pt adatoms on Pt(111). Such atomic manipulation has provided an elegant platform to study the confinement of electrons in nanoscale dimensions [7–9]. The simple quantum mechanical (Rae, 1992) model of electrons confined in a one-dimensional potential well of length, L (Fig. 11.3.7), also known as the “particle in a box” model, predicts that the energy levels of the electron, for an infinitely high potential, V0 → ∞, are discrete and given by En =

h2 n2 8me L 2

(11.3.2)

where n = 1, 2, 3, . . . . The corresponding wave functions are  ψn =

2 nπ x sin L L

(11.3.3)

for 0 < x < L, and ψ = 0, for regions outside the potential well. However, if V0 is finite, then the number, n, of wave functions within the potential well is finite; moreover, the probability for the wave function to extend outside the well is also finite with exponentially decreasing amplitude as a function of distance from the boundary. These fundamental ideas of quantum mechanics were spectacularly demonstrated by confinement of electrons in two-dimensional boxes, or quantum corrals, created by moving and positioning adsorbed atoms on surfaces. In one of the first such experiments [5, 7], 48 Fe atoms were positioned in the form of a circle on a Cu(111) surface (Fig. 11.3.8a). The itinerant electrons of Cu confined within this “circle” behave as though they are a two-dimensional electron gas and form a standing wave pattern. Moreover, STM measurements of the electronic structure in the quantum corral, are in good agreement with calculations (Fig. 11.3.8b).

OUP CORRECTED – FINAL

Physics of Scanning Force Microscopy 759 (b) 1.2 Height (Å)

(a)

1.0

48 Atom Ring 0.01V

Data Theory

0.8 0.6 –100

–50

0 Distance (Å)

50

100

Figure 11.3.8 Spatially resolved STM image of the eigenstates of a two-dimensional electron gas in a quantum corral. (a) 48-atom Fe ring, with average diameter ∼142.6 Å, constructed on Cu(111) enclosing a defect-free region of the surface. (b) Cross-section of the image data (solid line) and fit to theory (dotted line). Adapted from [7].

11.4 Physics of Scanning Force Microscopy The STM is a versatile imaging/spectroscopy tool, but it is restricted only to the study of conducting surfaces. Moreover, it was recognized very early in the development of the STM that whenever the tip–specimen distance was small enough for the flow of the tunneling current, significant forces would act simultaneously on the tip. It was speculated that if these atomic-range forces could be sensed, then an “atomic” force microscope, which can be applied to nonconducting materials, may be a possibility. This was indeed realized with the invention of the SFM [2] (Fig. 11.4.1). Here, the tip height is controlled such that the force between the sharp tip and the specimen during the scan is maintained at a constant value, allowing the measurement of the topography of the surface irrespective of its conducting properties. To measure the local force, the tip is mounted on a cantilever, which serves as a force sensor (§11.4.2). In principle, either the deflection of the cantilever or its dynamic response to the tip–specimen forces can be measured by a variety of methods. Strictly speaking, the total force combining the different contributions of the tip and the cantilever is measured (§11.4.3). Both short- and long-range forces should be considered; for example, the short-range van der Waals interaction is confined to the mesoscopic tip, but the long-range electrostatic interactions largely affect the cantilever. In addition to the contact mode described above, the SFM is also operated in two other modes: the dynamic mode, also known as noncontact AFM that provides true atomic resolution [10], and the tapping mode, also known as the intermittent contact mode that is widely used for imaging in air (§11.5). We now discuss these aspects of SFM and conclude this chapter by describing some applications of this versatile

OUP CORRECTED – FINAL

760 Scanning Probe Microscopy

Figure 11.4.1 The first atomic force microscope built by Binnig et al. [2], on display at the Science Museum, London. The original “cantilever” used was a strip of gold foil with a small diamond tip glued to it. Image downloaded sciencemuseum.org.uk.

from

http://

(b)

(a) A-B C-D w Figure 11.4.2 (a) A SFM tip and cantilever with important dimensions indicated. One common way to measure the displacement of the cantilever is to use an optical lever and a split photo-detector as shown. Both (b) normal forces, and (c) lateral forces can be measured. (a) Adapted from Meyer, Hug, and Bennewitz (2004). (b, c) Adapted from Colton et al. (1998).

l

FN (c)

t h z

x

y FL

technique in characterizing both physical and biological materials. Advanced readers should consult specialized books for more information, including Meyer, Hug, and Bennewitz (2004), and Eaton and West (2010).

11.4.1 Mechanical Characteristics of the Cantilever Unlike the STM, the central element of a SFM is the spring that measures the forces between the tip and surface. To measure normal tip–specimen forces it is important that the force sensor be rigid in two dimensions but relatively flexible or soft in its motion in the third dimension. A cantilever beam satisfies these requirements very well, and hence a microfabricated cantilever (Fig. 11.4.2) is

OUP CORRECTED – FINAL

Physics of Scanning Force Microscopy 761 normally used as the force detector. The force constant, kf , also known as stiffness, for normal bending of a cantilever is kf =

Ewt 3 4l 3

(11.4.1)

where E is the Young modulus5 of elasticity of the material, and w is the width, t is the thickness, and l is the length of the cantilever (Fig. 11.4.2a). The resonance frequency, f 0 , of this cantilever is 0.162t f0 = l2

E ρ

(11.4.2)

where ρ is its mass density. Note that by measuring the resonance frequency, as well as l and w, which can be obtained by SEM imaging, the thickness, t, of the cantilever can be determined. Cantilevers are commonly microfabricated of silicon; thus ρSi = 2330 kg/m3 , ESi = 169 GPa and f0Si =

t 1 l 2 7.23 × 10−4

(11.4.3)

The important properties of a SFM cantilever are its stiffness, kf , its resonance frequency, f 0 , and its variation with temperature, ∂f0 /∂T , its quality or Q factor,6 as well as its composition. For SFM applications, the resonance frequency should be significantly above ambient (building) vibrations (1–100 Hz), or sound frequencies (1–10 kHz). Further, the typical spring constant for atoms in a solid is katom ∼ matom ω2 , where matom ∼ 5 × 10−26 kg, and ω is of the order of phonon frequencies ( ∼ 1013 Hz). For optimal SFM performance, the stiffness, kf , of the cantilever should be of the same order of magnitude as katom . Typical values are kf ∼ 0.1–100 N/m. Further, to obtain atomic sensitivity, the thermal motion of the cantilever should be less than 1 Å. Thus, to avoid any thermal effects in SFM

 measurements, kf z2 ≥ kB T. At room temperature, this gives kf ≥ 0.4 N/m. To achieve such spring constants, we can see from (11.4.1) that the cantilevers must have dimensions in the micrometer range. Moreover, the resonance frequency (11.4.2) of a cantilever depends inversely on its physical size. If lateral forces are to be measured the torsional spring constant, kt , of the cantilever given by kt =

Gwt 3 3h2 l

(11.4.4)

where G is the shear modulus and h is the length of the tip, is important. In addition to the beam (Fig. 11.4.3b), triangular cantilever geometries (Fig. 11.4.3a) are widely used for topographic imaging. This is because their high

5 σ = Eε, where σ is the stress, ε is the strain, and E is the Young modulus in units of GPa (109 N/m2 ). 6 The Q factor defines the resonance behavior of an oscillator. Typically, it is defined as the ratio of the resonance frequency, f 0 , to its resonance width, f 0 , i.e. Q = f0 /f0 . The Q factor (a dimensionless quantity) depends on the damping mechanisms present in the cantilever; for cantilevers operated in air, Q is a few hundred, but in vacuum, Q reaches hundreds of thousands.

OUP CORRECTED – FINAL

762 Scanning Probe Microscopy torsional stiffness prevents them from being affected by lateral forces during the measurements. Details of the calculation of spring constants for triangular cantilevers can be found in [12]. Cantilevers are typically made of single crystalline silicon (beam) or silicon nitride (triangular) and manufactured by microfabrication processes in a clean room. In the case of silicon, the pyramidal tip points along and has a cone angle of ∼50◦ ; the cone angle is etched further to a tip radius of ∼10 nm. The silicon is highly doped to prevent charging effects. A silicon tip that is removed from the wafer has a coating of native silicon oxide, which is removed by etching with HF or by sputtering in a vacuum (Fig. 11.4.3c). The lateral resolution of an AFM can be improved by sharpening the tip, and many techniques are available to do so (Fig. 11.4.3d–g). See also Example 11.5.1. In the next section we discuss different approaches to monitor the motion of the cantilever as it functions as a deflection sensor of the tip.

(d)

(e)

(a)

(b)

50 μm

50 μm

(f)

(c) 50 Å

(g)

Figure 11.4.3 Scanning electron images of (a) triangular (Si3 N4 ) and (b) beam (Si) shaped cantilever probes. (c) HR-TEM images of a very sharp silicon tip with the native oxide removed by etching; the amorphous coating is polymerized hydrocarbon—notice the crystal structure remains true to the bulk even at the very tip. Different types of sharpened tips: (d) standard Si tip, (e) electrochemically etched super-sharp Si tip, (f) ion-milled, high aspect ratio Si tip, and (g) a tip with a carbon nanotube attached. (a–c) Adapted from [13]. (d–g) Adapted from Eaton and West (2010).

Example 11.4.1: A Si cantilever has the dimensions: l = 0.225 mm, w = 0.038 mm, t = 0.006 mm, and h = 0.125 mm. What is its bending force constant, torsional force constant, and resonance frequency? What is the maximum thermal motion of the tip at room temperature? Solution: For silicon, the Young modulus, E = 1.69 × 1011 N/m2 , Shear modulus, G = 0.68 × 1011 N/m, and density ρ = 2330 kg/m3 .

OUP CORRECTED – FINAL

Physics of Scanning Force Microscopy 763

Then, from (11.4.1), the bending force constant kf =

  3  1.69 × 1011 0.038 × 10−3 0.006 × 10−3 Ewt 3 = = 30.45 N/m.  3 4l 3 4 0.225 × 10−3

 The thermal motion of the tip z2 ≤

kB T kf

=

Thus (z) ≈ 1.14 × 10−11 m. From (11.4.3), the resonance frequency f0Si = From (11.4.4), the torsional spring constant kt =

Gwt 3 3h2 l

=



  3 0.68×1011 0.038×10−3 0.006×10−3 2    −3 −3 0.225×10 3 0.125×10

25×1.6×10−22 30.45 t 1 l 2 7.23×10−4

= 1.32 × 10−22

= 163.9 kHz.

= 52.92 N/m.

11.4.2 Cantilever as a Force Sensor In AFM/SFM it is important to measure the forces between a sharp tip and the specimen surface. This geometry results in very high pressure (force/area) on the sharp tip. In addition to preventing the tip from breaking, the AFM has to be optimized for measuring very weak forces. To accomplish this a number of force sensor designs (Fig. 11.4.4) have been developed and implemented for AFM/SFM. The most common method used in the majority of commercial instruments is the optical beam deflection method. In this design, a light beam is deflected from the back side of the cantilever and its deflection is monitored using a four-segment, position sensitive, photodiode detector (Fig. 11.4.2a). This widely used design can measure very small movements in both vertical deflection (Fig. 11.4.2b) and torsional rotation (Fig. 11.4.2c) of the cantilever, providing very high normal and lateral force sensitivity. Figure 11.4.4d shows an alternative way to detect the cantilever deflection by using the cantilever as the moving mirror in a Michelson interferometer (introduced earlier in Figure 3.5.1). This technique is very sensitive and is limited only by the thermal noise of the cantilever; moreover, potentially it can be implemented in the limited confines of an ultra-high vacuum (UHV) chamber and can also operate at low temperatures. However, this interferometry technique has met with only limited success because the probe can jump between interference fringes while scanning. In the first AFM built by Binnig et al. [2], an STM tip positioned on the back side of the cantilever (Fig. 11.4.4a) was used to measure the motion of the cantilever via the variations in the associated tunneling current. Even though this technique is a viable force sensor, its implementation is very difficult and it suffers the added complexity of accounting for the forces between the STM tip and the cantilever. Other methods integrate the sensor and the actuator (cantilever), creating significant advantages for dynamic measurements. It is also advantageous to

OUP CORRECTED – FINAL

764 Scanning Probe Microscopy

STM

Figure 11.4.4 Schematic representation of various deflection sensors designed for scanning force microscopy. Adapted from Colton et al. (1998).

(a) Electron tunneling

(b) Optical beam deflection

(c) Capacitance

(d) Interference

R R

R

(e) Piezoresistance

V (f) Piezoelectricity

implement such deflection sensors in UHV environments as they do not require additional positioning of the sensor elements in situ in the chamber. For example, the cantilever and a counter electrode can be microfabricated as an integral part of the force sensor, and the cantilever deflection can be monitored by any changes in the capacitance between them (Fig. 11.4.4c). Such a technique is capable of fast measurements but the forces between the cantilever and the counter electrode may affect the performance. In dynamic AFM, it was recognized quite early in the development of the technique that the requirements for the force sensor were similar to that of the time-keeping elements (quartz tuning forks) manufactured in the billions of numbers for low-cost watches. In principle, these tuning forks are cheap piezoelectric sensors and by attaching metallic tips to one of the prongs of the tuning fork (while keeping the other one fixed) they were modified to serve as AFM force sensors (Fig. 11.4.4f). Alternatively, high-quality self-sensing force sensors with a piezoresistive element integrated into the back side of the cantilever

OUP CORRECTED – FINAL

Physics of Scanning Force Microscopy 765 have been developed (Fig. 11.4.4e). Here, the resistance changes with the bending of the cantilever and it can be monitored as a force sensor.

11.4.3 Tip–Specimen Forces Encountered in an SFM Except for the force sensor replacing the tunneling tip, the SFM is not conceptually different from an STM. The component of the tip–specimen force, Ft-s , in a direction normal to the surface that is relevant for SFM, has both short- and longrange contributions. In principle, for an SFM operating in a vacuum to achieve atomic resolution, the short-range (< 1 nm) chemical forces must dominate over the van der Waals, and the long-range ( 5 nm. To first order, approximating the tip–specimen geometry as a hemisphere (tip) in the proximity of a semi-infinite body (specimen), the van der Waals force (Israelichivli, 1991) is given by FvdW =

AH R 2 6dt-s

(11.4.5)

where AH is the material-dependent Hamaker constant [27], R is the tip radius, and dt-s (∼0.5 nm) is the tip–specimen separation. The Hamaker constant is of the order of 10–19 J, with an enhancement sometimes as large as 30 times for metals compared to insulators. For typical SFM conditions, we get FvdW ∼ 2 nN for commercial silicon tips and FvdW ∼ 7 − 10 nN for etched metal tips with R ∼ 100 nm. This magnitude of FvdW is a significant hindrance to achieving atomic resolution in scanning force microscopy. However, FvdW can be significantly reduced by immersing the cantilever in water [14]. In addition, electrostatic forces, Fel , become important if the specimen and tip are both conductors and have an electrostatic potential difference, U . Here, U = Ubias −Ucpd , typically ∼1 V, where Ubias is the tip–specimen bias voltage and Ucpd is the contact potential difference arising from the different work functions of the tip and specimen. We can then show [15] that Fel =

π ε0 R (U )2 dt-s

(11.4.6)

with Fel ∼1.6 nN for standard tip and specimen parameters; here ε0 = 8.854 × 10−12 Fm−1 is the permittivity of free space. Alternatively, if the tip and specimen are ferromagnetic, the tip will experience a magnetic force, Fmag , given by  Fmag = −μ0

  ∂   Hsample + t dV Mtip x , y , z ∂z

(11.4.7)

  where x , y , z is a coordinate frame attached to the tip, t (x,y,z) is the position of the tip in the laboratory frame of reference, Mtip is the magnetization of the tip, Hspecimen is the field arising from the specimen, and the integral is over the volume, V , of the magnetic tip. We discuss magnetic force microscopy (MFM), using ferromagnetically coated tips in §11.8.2. Finally, in ambient conditions, capillary forces, Fcap , arising from water vapor condensation, especially if the tip radius, R < 100 nm, have to be considered. The water molecules condensing between the tip and specimen will from a meniscus (Fig. 11.4.5). The capillary force can be estimated as Fcap =

4π Rγ cos θ 1+

dt-s R(1−cos φ)

(11.4.8)

OUP CORRECTED – FINAL

Operation of the Scanning Force Microscope 767 where γ is the surface tension of water (∼0.074 N/m, at 20 ◦ C), θ is the contact angle, and φ is the angle of the meniscus. Now the maximum value of the capillary force is Fcap = 4πRγ cos θ, which is significantly larger (∼90 nN) than the corresponding van der Waals force. The effect of the capillary force can be substantially reduced, if not eliminated, by coating the tip and specimen surface with hydrophobic amphiphilic molecules. As mentioned earlier, immersing the cantilever in water also can eliminate the capillary forces entirely. On a positive note, capillary forces can be exploited for nanoscale lithography, and §11.8.5 discusses the related technique of dip-pen nanolithography (DPN).

Tip θ φ

Example 11.4.2: Calculate the magnitudes of the van der Waals force for commercial silicon and etched metal tips. Also, calculate the electrostatic and the maximum capillary forces for typical scanning probe parameters. Solution: For van der Waals force, we apply (11.4.5) with the following parameters: AH = 10−19 J, R ∼ 30 nm and dt-s ∼ 0.5 nm for Si, with the radius R∼100 nm for etched metals. Thus, we get FvdW = 2 nN for Si, and FvdW = 6.67 nN for etched metals. To calculate the electrostatic forces, we apply (11.4.6) with R ∼ 30 nm, dt-s ∼ 0.5 nm, and U ∼ 1 V, and get Fel = 1.67 nN. The maximum capillary force, for R ∼ 100 nm and dt-s ∼ 0.5 nm, is given by the numerator of (11.4.8), i.e. Fcap = 4π Rγ cos θ ∼ 90 nN.

11.5 Operation of the Scanning Force Microscope The wide range of operating modes developed in scanning force microscopy [16], can be broadly classified as static and dynamic, where the bending (former) of the cantilever, or changes in vibrational properties (latter) are measured. In the static mode, the cantilever tip is in “contact” with the surface, and the tip–specimen distance is controlled to maintain a constant bending of the cantilever. Thus, the topography of the specimen, subject to a constant force on the cantilever, is recorded. Alternatively, the tip in “contact” is scanned at a constant distance from the surface and the variation in the force recorded as the image. In contrast, the dynamic mode measures changes in the vibrational properties of the cantilever due to the local tip–specimen interactions. These properties include its resonance frequency, the amplitude of oscillation, and the phase difference between the induced excitation and resultant oscillations of the cantilever. Further, dynamic modes are differentiated based on feedback parameters, i.e. amplitude, frequency, or phase modulation, which are used to sense or control the tip–specimen distance [17]. Dynamic modes include both contact and noncontact

R

dt-s

Meniscus

Surface Figure 11.4.5 Schematic representation of the formation of the meniscus between the tip and specimen by the condensation of water vapor, which leads to capillary forces.

OUP CORRECTED – FINAL

768 Scanning Probe Microscopy modes of operation; however, the most common method is the intermittent contact or tapping mode of operation. In addition to topography, the tapping mode provides information on the physical and chemical properties of the surface while causing minimal damage during the scanning. Note that SFM modes can provide information beyond surface topography and properties by recording various signals (Fig. 11.1.1) as a function of distance or other parameters. Such measurements, known as force spectroscopy, provide wide-ranging information on tip–specimen interactions. It is important to understand the difference between the interaction forces in noncontact and contact modes of operation. The tip–specimen interaction in the noncontact mode (Fig. 11.5.1a) consists of attractive long- and shortrange contributions, with the latter making a significant contribution to the total force. Because the short-range forces between the outermost atoms of the tip and specimen are significant, atomic resolution imaging in noncontact mode Dynamic

(a)

Static

(b) Contact Region

Contact Region

Tip

Tip

Specimen

Specimen (d)

(c)

Total

zd

A

Force

Force

Short Range

Distance

zs

F/k

Figure 11.5.1 Interaction forces and contrast formation in (a, c) dynamic (noncontact), and (b, d) static (contact) modes in SFM. In (a) and (b) the arrows indicate the direction of forces on the atoms of the tip and specimen; note that the contact region, highlighted in grey, laterally extends over a region of several atoms in (b). The corresponding force-distance curves, including the total (continuous) and short-range (dashed) forces, are shown for (c) noncontact, and (d) contact modes. In (c), the amplitude range, A, of the oscillating tip used in dynamic measurements is indicated. In (d), arrows indicate the points where the tip jumps into and out of contact, with the slope of the straight dotted line indicating the stiffness, k, of the cantilever. Adapted from Meyer, Hug, and Bennewitz (2004).

OUP CORRECTED – FINAL

Operation of the Scanning Force Microscope 769 is possible without any surface damage. Alternatively, in the contact mode the situation is complex as some of the atoms of the tip and specimen are in repulsive contact (Fig. 11.5.1b). Even though the total force, Ft-s , may be attractive, the local deformation of the surface and the transition between attractive and repulsive forces may lead to a contact region that spans several atoms; as such, atomic resolution in contact mode is not possible. Static, contact mode topographic imaging is discussed in §11.5.1. To further get an idea of the static and dynamic modes of operation, we have to consider the force–distance behavior as the tip is brought close to surface. Note that, typically, static and dynamic modes use cantilevers that are mechanically soft and rigid, respectively. In the dynamic mode (Fig. 11.5.1c), the position, zd , in the force–distance curve corresponding to Figure 11.5.1a indicates a situation where the short-range attractive forces are a maximum and form a significant component of the total force. Further, the amplitude range, A, of the oscillations of the tip is always maintained such that the closest distance of approach of the tip to the surface prevents it from being trapped at the specimen surface. In this scenario, as shown, the total force is attractive even when the short-range force becomes repulsive. Dynamic, noncontact SFM imaging is discussed further in §11.5.3. The force–distance curve for the contact mode (Fig. 11.5.1d) indicates a position, zs , for the tip that is on the repulsive side of the short-range force curve. Further, as the tip is brought close to the surface in the static mode, it is characterized by an instability (indicated by the left arrow in the figure) known as “jump to contact,” which arises from the elastic bending of the cantilever. Similarly, as the tip is retracted from the surface, it suffers another instability (indicated by the right arrow), known as “jump off contact.” These instabilities can be understood with reference to the slope of the dashed line in the figure, which represents the spring constant, k, of the cantilever. Approaching the specimen, when the derivative of the total force curve becomes larger than k, the tip jumps to a contact position where the short-range repulsive force balances the attractive force. Similarly, upon retraction of the tip, the jump off occurs when the total force derivative becomes equal to the spring constant. We now discuss some important modes of SFM operation in more detail.

11.5.1 Static Contact Mode for Topographic Imaging The very first mode developed for SFM, static contact mode uses static measurements of the deflection of the cantilever to create a topographic image of the surface. In this mode, the tip is first brought into contact with the surface and its height adjusted to be in the short-range repulsive force regime (Fig. 11.5.1b). Overall, the tip position defines a force equilibrium between the longrange attractive force, Fatt , of the tip and surface, and the short-range repulsive force, Frep , of the tip apex and specimen, and external force, Fext , due to the cantilever stiffness (Fig. 11.5.2a). Strictly speaking, in this regime, a combination

OUP CORRECTED – FINAL

770 Scanning Probe Microscopy of cantilever bending and local specimen compression will occur, and the set point height is chosen to minimize the resulting force and any related damage to the specimen surface or the tip but, at the same time, to avoid jump out-of-contact of the tip. Once the vertical position of contact is established, the topographic image can be obtained by operating the scan in two different ways (Fig. 11.5.2b). In the constant force method, also known as the constant deflection method, the microscope feedback system is used to keep the cantilever deflection at a set value determined by the operator. From Hooke’s law, the force, F, applied by the probe to the surface is F = −k × D, where k is the spring constant of the cantilever and D is its vertical displacement. As a result, a probe with a small spring constant (a soft cantilever) will maximize the vertical deflection for a given force. As the tip (or specimen) is scanned, the short-range repulsive forces are very sensitive to the tip–specimen distance, and hence, images of constant repulsive forces can be interpreted in terms of the surface topography. Alternatively (Fig. 11.5.2b, bottom), the feedback system can be turned off and an image, scanned at constant height, can be recorded. Now the image comes entirely from the deflection of the cantilever and not the voltage applied to the z-piezo drive. Hence, this measurement is dependent on the specific calibration of the cantilever deflection. The resolution of the contact mode SFM, depends on the lateral extent of the contact region (Fig. 11.5.1b) of the tip-apex with the specimen. The contact diameter, dc , assumed to increase with the applied force, F, due to the elastic deformation of the tip and specimen (so-called Hertz model), is given by

(a)

(b) Fext = kΔz

Constant force mode

Frep Fatt Specimen

Constant height mode

Figure 11.5.2 (a) Force balance affecting the cantilever displacement in the contact mode include the short-range tip-apex-surface repulsion, Frep , the longrange tip–specimen attractive force, Fatt , and the external force, Fext , due to the cantilever spring constant. (b) Two different scanning modes are used in the contact mode for topographic imaging: the constant force or constant displacement mode (top) plotting the voltage applied to the feedback system to keep the cantilever deflection at a set value, and the constant height mode (bottom) where the feedback loop is turned off and the imaging signal is based on the cantilever deflection rather than the voltage applied to the z-piezo drive.

OUP CORRECTED – FINAL

Operation of the Scanning Force Microscope 771 dc = 2(DRF)1/3

(11.5.1)

    where R is the tip radius, D = 1 − νt2 /Et + 1 − νs2 /Es , and ν and E are the Poison ratio and Young modulus of the tip (t) and specimen (s), respectively. Again, for typical values of R∼90 nm, Et = Es = 1.5 × 1011 N/m2 , νt = νs = 0.3, we get dc ∼2–10 nm, for F∼1–100 nN, which is valid for ambient atmospheres. Thus, these contact diameters limit lateral resolution in contact SFM. However, if the force can be reduced, such as in high vacuum, to 0.1–10 nN, we can expect a smaller contact diameter (dc ∼1–4 nm). Alternatively, in liquids where the longrange force can be significantly reduced to ∼10 pN, atomic resolution may be possible. Note that in this mode the tip is always in contact with the surface, and the ensuing short-range repulsive force may damage either the specimen surface, or the tip, during the scanning process. Since the tip is always in contact, in addition to the normal force, the tip and/or specimen may also experience lateral forces. Last, but not least, since the nature of the specimen surface may affect the result, this method could also probe the local nature/properties of the specimen. Example 11.5.1: An alternative model, adapted from [46], of a contact mode operation of the AFM is shown in the following figure. The cantilever is represented by a spring of force constant, k, and the tip is represented as a sphere of radius, R (typically ∼50 nm), and in the absence of external forces is separated by a distance, Z, from the specimen surface. The forces acting on the tip cause a deflection, δ (negative), from equilibrium, such that the actual separation is D = Z + δ, as shown in (a). (a)

(b) Spring, k R Tip

δ Z

a

D=Z+δ (i) Neglecting short-range forces, the equilibrium position satisfies the condition given by AH R 2 + kδ = 0, where AH is the Hamaker 6(Z+δ) constant. This equation can be scaled such that it depends only on one parameter, l = (AH R/k)1/3 . Further, at a distance, Dcrit = l/31/3 , the tip becomes unstable and snaps into contact with the surface as shown in (b). Calculate the scaling parameter, l, and the critical distance, Dcrit , for typical values of tip parameters.

OUP CORRECTED – FINAL

772 Scanning Probe Microscopy

(ii) When the tip snaps into contact, Figure (b), assume that only the specimen is deformed. Then, the characteristic resolution scale will be defined by the region of contact which will have a minimum  1   3 2 R 2 AH radius, R = where K1 = 32 1−υ , and the minimum 2 E 8KDmin

˚ Calculate distance between the sphere and surface is Dmin = 1.5 A. R for typical tip parameters. How can the resolution be improved? Solution: Typical parameters for the tip are E = 1.5 × 1011N/m2 , R = 1−0.32 50 nm , υ = 0.3, k = 0.4 N/m and AH = 10−19 J. Thus K1 = 32 1.5×10 = 11 9.1 × 10−12 .

(i) The scaling parameter l = (AH R/k)1/3 = 3.45 nm and the critical distance Dcrit = l/31/3 = 2.4 nm.  1/3  2 50×10−9 10−19 (ii) The minimum radius of contact R =  =    2 11 −9 8 1.1×10

0.15×10

2.84 nm, which will be the best achievable resolution. There are two ways to improve the resolution: 1) by reducing the radius, R, of the tip, either by developing “super tips” of smaller radius (∼10 nm) or by sharpening existing tips, and 2) by reducing the Hamaker constant, by immersing the tip and specimen in different liquids.

11.5.2 Lateral Force Microscopy In contact mode, the vertical deflection of the cantilever, indicated by the difference in the signal between the top and bottom half of the split photodiode detector (Fig. 11.4.2b) is used as the feedback signal. In addition, it is possible to compare the signals from the left and right sides of the split photodiode (Fig. 11.4.2c). If we do the latter, we will obtain information about the lateral deflection of the cantilever, which can be correlated with the mechanical interaction, particularly friction, of the probe with the specimen surface. Thus, this technique is referred to as lateral force microscopy (LFM) or as friction force microscopy (FFM). Strictly speaking, the lateral bending and the vertical deflection of the cantilever are coupled, and thus such scans provide information on both the specimen topography and the materials properties; in particular, the friction depends on the slope of travel of the tip. The lateral deflection is always different in two different directions (e.g. forward and backward scans; Fig. 11.5.3) Even on perfectly flat and homogeneous specimens, the two signals will differ in magnitude. In general, changes in slope will affect forward and backward lateral scans in opposite fashion, and changes in materials friction properties will exhibit greater or smaller difference in forward and backward scans.

OUP CORRECTED – FINAL

Operation of the Scanning Force Microscope 773 Orientation of the probe

Lateral deflection signal

Topography Change

Scan direction

Specimen Scan direction

Specimen

Materials Friction

Scan direction

Specimen

Scan direction

Specimen Figure 11.5.3 Lateral signal recorded from a specimen with only (top) topographical, and (bottom) materials friction variations. In the lower figure, darker colors represent higher friction. If the lateral deflection signal for the forward scans is subtracted from the reverse scan, it will show a constant value for the topography changes. However, such subtraction for materials friction, will give a significant difference variation and is indicative of local properties. Adapted from Eaton and West (2010).

11.5.3 Dynamic Noncontact Modes of Atomic Force Microscopy In dynamic mode, the amplitude, the resonance frequency, and the phase shift of the oscillation link the dynamics of the vibrating cantilever with a sharp tip to the local tip–surface interactions. In fact, any of the three parameters can be used

OUP CORRECTED – FINAL

774 Scanning Probe Microscopy as the feedback parameter to locally sense and image the topography and other properties of the surface. In the amplitude-modulated atomic force microscopy (AM-AFM), a stiff cantilever with a sharp tip is mounted on an actuator and externally excited to oscillate at a fixed amplitude, Adrive , and a drive frequency, fdrive , that is close to but different from its resonance frequency, f0 , given by (11.4.2). As the tip approaches and interacts with the specimen, it causes a change in both its amplitude and phase (with respect to the driving signal) of oscillation. The oscillation amplitude is used as the feedback parameter to measure the topography. In addition, the phase shift variation could be used to map the variation in materials properties. Unfortunately, the change in amplitude as a result of tip–specimen interaction only occurs over a time scale of τAM = 2Q/f0 , where Q is the quality factor. In vacuum, Q factors can be quite large (∼105 ) and the AM-AFM mode can be rather slow. To overcome this, the frequency modulated AFM (FM-AFM) was invented (Fig. 11.5.4). Tapping or intermittent-contact mode (IC-AFM) is a dynamic mode that allows the probe to touch the specimen briefly to experience a repulsive tip– specimen interaction. In this mode the feedback is usually based on the amplitude modulation. Moreover, unlike in contact mode where the lateral forces cause problems, here they are eliminated as the tip moves perpendicular to the specimen as it is scanned. Further, since the cantilever withdraws the tip away from the specimen, any capillary forces from water vapor or other contamination on

Amplitude

f

(1) Frequency shift Intermittent Damping A (2)

fdrive Frequency

Frequency

Figure 11.5.4 Two dynamic non-contact modes are used to detect the tip– specimen interaction in atomic force microscopy. (a) In frequency modulation (FM-AFM), the shift in frequency, f, when the cantilever is driven at its resonance frequency, caused by the tip–specimen interaction is detected. (b) In amplitude modulation (AM-AFM), the change in amplitude, A, at a fixed frequency is detected. One variant of AM-AFM is the intermittent contact (IC-AFM) or tapping mode. In this case, the decrease in amplitude, A, can arise from a frequency shift (1), or the damping (2), when the tip touches the surface. Adapted from Meyer, Hug, and Bennewitz (2004).

OUP CORRECTED – FINAL

Operation of the Scanning Force Microscope 775 the specimen surface affects it minimally. In IC-AFM, both the change in the amplitude as well as the phase (delay) of the probe is recorded (Fig. 11.5.5). The latter is sensitive to materials properties (Fig. 11.5.6). AM-AFM imaging also provides the opportunity to map variations in the composition, friction, viscoelasticity, and adhesion properties of the specimen surface. The method is called phase contrast imaging and it measures the phase lag of the tip with respect to the drive signal, while the feedback keeps the amplitude constant. Figure 11.5.6 illustrates the difference in contrast obtained by standard topography and phase contrast obtained in AM-AFM imaging of a thin film of hydrogenated diblock copolymer (PEO-PB). The topography image shows no specific features; however, the phase contrast image shows that the copolymers

0°

180° Intermittent Contact Oscillation

Phase Shift Free Oscillation A

Figure 11.5.5 The effect of intermittent contact (tapping mode) on the oscillation of the cantilever includes a reduction in amplitude, A, due to the repulsive contact with the specimen surface and a phase shift determined by the properties of the material. (a)

(b)

Figure 11.5.6 AM-AFM images of a block copolymer mesophase. (a) Topography, and (b) phase contrast image. In (b) individual spheres (12 nm diameter), which are either crystalline (light) or non-crystalline (dark) PEO micelles, are resolved. The long edge of the image is 1 mm and the maximum height variation detected in (a) was 10 nm. Adapted from [17].

OUP CORRECTED – FINAL

776 Scanning Probe Microscopy organize into spherical PEO micelles with crystallization occurring individually for each sphere. Alternatively, in FM-AFM the cantilever is kept oscillating at its resonance frequency with a fixed amplitude. The local force gradient, F , between the tip and specimen causes a change in the effective cantilever stiffness, i.e. k = F . Now, the fractional change, f, in the cantilever resonance frequency, f0 , can be related to the fractional change in the cantilever stiffness as f k F = = f0 2k 2k

(11.5.2)

The spatial dependence of the frequency shift, f, is the source of contrast in the FM-AFM image. The important advantage of this method over AM-AFM is that the change in resonance frequency occurs very fast within a single oscillation over a time scale τFM = 1/f0 . Initially, both AM-AFM and FM-AFM were conceived to be noncontact modes of operation and the cantilever was maintained at sufficient distance from the specimen to obtain a net attractive force between the tip and specimen. Note that most AFM experiments in UHV are performed in FM mode, whereas experiments in air or in liquids are performed in the AM mode.

11.6 Scanning Force Microscopy Instrumentation The principal operation of an SFM is shown in the block diagram of Figure 11.6.1a, and the components of the stage, which is the heart of the instrument, are highlighted in Figure 11.6.1b. The stage includes the specimen holder, X-Y specimen positioning platform, the coarse vertical approach mechanism (Z motor), the specimen x-y-z scanner, and usually, an optical microscope to help position the tip over the microstructural feature of interest in the specimen. The propensity of the stage to vibrate mechanically varies inversely as its physical size, and for high resolution it is constructed to be as small as possible. In addition, the single most important factor that influences the vertical resolution of an AFM is the rigidity of the mechanical loop of the instrument. By the mechanical loop, we mean all the elements (x-y-z scanner, X-Y specimen stage, Z motor, and the probe) that are required to hold the probe at a precise and fixed distance from the surface. The x-y-z scanners, which control the fine movement of the probe, are normally constructed of amorphous lead–barium–titanate (PBT) or lead–zirconium– titanate (PZT) piezoelectric ceramics materials and are optimized either for maximum expansion with respect to the applied voltage or for linear expansionvoltage characteristics. Generally, one of the two factors mentioned is optimized at the expense of the other. A piezoelectric ceramic preserves its volume as it changes its geometry on the application of an electric voltage. Based on this simple idea, two standard

OUP CORRECTED – FINAL

Artifacts in Scanning Probe Microscopy 777 (a)

(b) x-y piezo

Video microscope lens

X-Y raster electronics

z piezo Force transducer

Compare

Probe Set force Specimen

Feedback controller

x-y-z probe scanner and force sensor

Specimen holder Image X-Y specimen stage

Z motors

Figure 11.6.1 (a) Block diagram of the AFM in operation. (b) A photograph of an AFM stage with principal components identified. Adapted from Eaton and West (2010).

configurations of piezoelectric scanners are designed to move the probe (or specimen) accurately in an AFM. The widely used tube scanner (Fig. 11.6.2a), with inner and outer electrodes, is compact, easy to fabricate, and produces precise movements, especially for small scan ranges. In addition, the clear optical path through the center of the tube makes for easier probe positioning; however, the scans suffer from nonlinearity. Alternatively, the tripod scanner (Fig. 11.6.2b) is the simplest three-dimensional scanner design and allows for independent control of motion of all three axes. In practice, all piezoelectric drives suffer from nonlinear behavior of hysteresis and creep as a function of applied voltage (see §11.7.2). In addition, an AFM/SFM instrument requires force sensors (the cantilever arrangements discussed in §11.4.2), appropriate electronics for generating voltage ramps corresponding to the x-y positioning of the probe, feedback control to drive the z-piezoelectric ceramic, and appropriate signal collection for amplitude, frequency, or phase modulation; these are discussed in detail in Eaton and West (2004). Finally, from a practical point of view, the image profile will depend on the shape of the probe, with a sharper probe often giving a better horizontal resolution, and many techniques (Fig. 11.4.3d–g) are used to obtain sharp probe tips.

11.7 Artifacts in Scanning Probe Microscopy 11.7.1 Probe Artifacts As mentioned in Figure 11.3.2, SPM images are a convolution of the specimen topography with the shape of the tip of the probe. In simple terms, all specimen features with a radius of curvature smaller than that of the probing tip are not properly imaged (Fig. 11.7.1a). If the probe tip is blunt, images of convex features,

OUP CORRECTED – FINAL

778 Scanning Probe Microscopy (a)

(b) y piezo

z l

–y

–x +y

+x

t +x

+y

Rigid Support Structure

x piezo z piezo

z –y

–x

Figure 11.6.2 The two types of piezoelectric scanners used commonly in AFM. (a) The tube scanner, with length, l, and thickness, t, moves in the x-y-z directions using four electrodes outside for x-y motion and an inner electrode for z-motion. The movement of the piezodrives are x = 2 y ∝ lt V and z ∝ tl V . (b) The simplest tripod scanner with x, y, z ∝ V . In both cases, V is the applied voltage. Adapted from Eaton and West (2010).

such as particles, would be larger than expected in the lateral dimension (Fig. 11.7.1b); on the other hand, concave features, such as holes on a flat surface, may appear narrower than they actually are (Fig. 11.7.1c). Moreover, if the features have a higher aspect ratio than the probing tip, they are also not imaged properly and instead produce repeating images of an inverted probe or its side walls (Fig. 11.7.1d). In addition, probes often get contaminated, and if such “dirty” probes are used they show repeating patterns in the images (Fig. 11.7.2a). Sometimes, when the tip is broken it can end up having a double tip; in this case, each feature is doubled and shows a false “twin” in the image (Fig. 11.7.2b). If this were to happen, it is best to replace with a fresh probe and check its performance using resolution standards available from national metrology laboratories or commercial SPM vendors. Example 11.7.1: For a probe geometry with cross-section described as an upside-down triangle of semi-angle, θ 0 , the diameter of a particle measured, dmeas , is related to its actual diameter, d, by a convolution term, i.e.       tan θ0 2 2 dmeas = d cos θ0 + cos θ0 + (1 + sin θ0 ) −1 + cos θ0 + tan θ0 Calculate the actual diameter for a particle with dmeas = 100 nm for (a) θ0 = 30

◦

OUP CORRECTED – FINAL

Artifacts in Scanning Probe Microscopy 779 (a)

Rt Figure 11.7.1 Probe tip artifacts: (a) Image of a standard surface of asperities with radius of curvature, Rs , much smaller than the tip radius, Rt . The profile measured (dotted line) is not representative of the surface but corresponds to an inverted probe tip of radius Rt . (b) Images (dotted lines) of a convex surface (particles) obtained by using two different tip radii. The height is accurate but the feature appears much broader when a probe of larger radius is used. (c) Images (dotted lines) of a concave surface (holes) appear less wide and less deep with the larger probe. (d) Images of features with high aspect ratio resemble images of the inverted probe (left) or its side walls (right).

(b)

(c)

(d)

(a) Adapted from Meyer, Hug, and Bennewitz (2004). (b–d) Adapted from Eaton and West (2010).

(a)

(b)

Figure 11.7.2 Examples of tip artifacts (a) Repeating images due to dirty or broken tips. (b) Images of DNA molecule with a broken tip with each molecule showing a false “twin” image next to it. 500 nm

Adapted from Eaton and West (2010).

OUP CORRECTED – FINAL

780 Scanning Probe Microscopy ◦

(b) θ0 = 60 . Solution: The convolution term for (a) is 1.732. Thus d = 100/1.732 = 57.7 nm (b) is 3.732. Thus d = 100/3.732 = 26.8 nm. Clearly, the sharper tip (a) requires a smaller convolution correction, but it is still substantial.

11.7.2 Instrument Artifacts The two major artifacts of piezoelectric scanners are hysteresis and creep. Piezoelectric materials exhibit hysteresis in the displacement versus voltage behavior, and when a saw tooth voltage is applied, the position of the actuator can deviate by as much as 15% between forward and backward movements. In addition, when an instantaneous voltage is applied and maintained, the piezoelectric material may not only move to the new position, but may continue to move in the same direction even when the voltage remains unchanged. This movement is called creep. Even though the duration of creep is short, it distorts the scanned image; in practice, by waiting long enough for the piezo position to stabilize before imaging, this artifact can be minimized. Alternatively, position sensors have been integrated in the cantilever force sensor to measure the true motion of the tip relative to the specimen, and make real-time scan corrections [18]. Unfortunately, such hardware correction systems offer poor signal to noise, making it difficult to achieve high resolution. In that case, software corrections that are based on empirical approximations are used. In addition, high-resolution AFM scanners also suffer from specimen drift. To avoid this artifact, specimens are often glued to the specimen support. Even then, thermal expansion of the specimen can appear as a movement of the specimen. By changing the scan direction, the nature of the drift can be clarified. In addition, by repeated measurement of a specimen feature the drift velocity can be determined; typically, this can be as much as nm/s at room temperature, but can be reduced by one order of magnitude to Å/s by lowering the operating temperature.

11.8 Select Applications of Scanning Force Microscopy SFM or AFM is now well established and successfully applied in many fields. We now review some representative examples involving high-resolution AFM imag-

OUP CORRECTED – FINAL

Select Applications of Scanning Force Microscopy 781 ing (§11.8.1), measurement of local magnetic (§11.8.2) and thermal (§11.8.3) properties of materials, applications in the life sciences (§11.8.4), and lithography on the nanoscale (§11.8.5). A comprehensive discussion of all aspects of the subject can be found in more specialized textbooks, including Meyer, Hug, and Bennewitz (2004) and Eaton and West (2010).

11.8.1 Atomic Fingerprinting in Frequency Modulated Atomic Force Microscopy As mentioned earlier, under UHV conditions, with excellent vibration isolation, and noncontact operation, FM-AFM can be used to obtain atomic resolution imaging. However, its development as a true metrology tool that is capable of chemically identifying individual atoms on the surface requires true innovation and careful calibration of the local chemical forces [20]. Recall that in FM-AFM, the atomic-scale interaction of the apex of the tip with the specimen surface atoms (Fig. 11.8.1a) involving short-range chemical and long-range van der Waals forces leads to a change in its resonant frequency. Such variation is measured as a function of the tip position and plotted as a three-dimensional atomic image of the surface (Fig. 11.8.1b). Normally, it is difficult to establish the chemical identity of the individual atoms on the surface, as the forces measured above individual atoms (such as Si, Sn, and Pb, in the surface alloy shown in Figure 11.8.1) vary from scan to scan and from tip to tip. This is understandable as the precise atomic shape and composition of the apex of the tip (Fig. 11.8.1a) may change over time and differ from tip to tip. Even though the atomic structure of the tip is impossible to determine and control reproducibly, Sugitomo et al. [20] discovered a clever way to address the problem and demonstrated the possibility of atomic fingerprinting in FM-AFM imaging. By repeatedly measuring a large number of force-displacement curves at atomic resolution for the three species (Si, Sn, and Pb), they observed that the strongest interaction was always between the tip and Si atoms. So, they normalized the maximum attractive force for individual atoms by the maximum force measured for Si atoms, and in this way, they created a distinctive fingerprint for the three different atoms, i.e. Si ∼100%, Sn ∼77%, and Pb ∼55% (Fig. 11.8.1c). Furthermore, using atomistic modeling [21, 22] of the interaction between the tip and surface, they developed a standard procedure [23] now used to interpret atomic-scale AFM imaging. First, they confirmed plausible atomic structures for the tip and then calculated the forces for the three different species that matched well with the experimental variation in their individual force curves. Now, armed with such fingerprinting, they used the measurement of the chemical forces as the basis of atomic recognition in the FM-AFM images of a surface alloy of Si, Sn, and Pb atoms (Fig. 11.8.1d).

OUP CORRECTED – FINAL

782 Scanning Probe Microscopy (a)

(b)

Short-range Chemical

–1

(d)

(c)

Van der Waal –2 Total

–3 1

2

3

4

5

6

7

8

9

Tip-surface distance (Å)

Atomic counts

Force (pN)

0

100 %

6 5 4 3 2 1

77 % 59 % Pb Sn Si

0.0 0.4 0.8 1.2 1.6 2.0 2.4 Maximum attractive total force (nN) Figure 11.8.1 Atomic fingerprinting with frequency modulated atomic force microscopy. (a) Operation of the microscope in dynamic mode, illustrating the onset of the chemical bond between the atoms at the apex of the tip and the surface. The inset shows the contributions to the force-distance curves from the chemical and van der Waals forces. (b) The atomic resolution dynamic mode image of the surface obtained without resolving the chemical species. (c) Calibration of the attractive force for the three atomic species, normalized with that of Si. (d) The same image as in (b) but now color coded to identify the local chemical composition at atomic resolution. Adapted from [20].

11.8.2 Magnetic Force Microscopy (MFM) As mentioned earlier (11.4.7), the spatial variation of the stray magnetic fields on the surface of a specimen can be measured directly in an SFM, provided the tip is coated with a ferromagnetic thin film. In practice, this can be easily accomplished by coating a standard silicon tip with a thin magnetic film; typically, Co or alloys of Co–Cr, Co–Ni, etc., are used. This may have two detrimental consequences: one, it increases the radius of the tip leading to a deterioration of the resolution; and two, it may possibly soften the surface of the tip, mechanically speaking, leading to an increased wear rate. Moreover, when the tip and specimen are in contact, the magnetic forces are one order of magnitude smaller than the other tip–specimen forces, which are dominated by the shorter-range van der Waals interactions. However, magnetic forces are reasonably long range, and hence it is

OUP CORRECTED – FINAL

Select Applications of Scanning Force Microscopy 783 Second scan follows topography and detects magnetic signal First scan in region of strong van der Waals forces

Surface topography (exaggerated for clarity)

Magnetized regions Measured topography signal from first scan Measured magnetic (MFM) signal from second scan

advantageous to measure them with the tip positioned 5–15 nm distant from the surface, to avoid interference from the other forces. The key to successful MFM imaging is to separate the magnetic signal from the total signal, which includes the topographical variation (Fig. 11.8.2). In practice, the separation can be achieved by carrying out two scans. In the first scan, the tip is positioned close to the surface, where the van der Waals image dominates, to obtain a topographic image. Then, for the second scan, the tip–specimen distance is increased to sense only the magnetic forces, but the set point based on the first scan is varied so that the tip follows the topography of the surface (Fig. 11.8.3). Now, magnetic contrast in the MFM image can be obtained by measuring either the amplitude, frequency, or phase change in the cantilever oscillations as a function of position. Alternatively, similar information can be obtained by intermittent contact, or tapping mode AFM, where the momentary contact of the tip with the surface monitors the surface topography and the noncontact portion of the scan monitors the magnetic forces (Fig. 11.8.3). MFM is reviewed in [24]. The technique compares very well with other modes of magnetic imaging; for further information, see Krishnan (2016). The spatial resolution in MFM depends on the magnetized part of the probe and the tip–specimen distance. Best lateral resolution can be achieved by keeping the magnetized part of the tip to the smallest volume possible, consistent with a detectable magnetic signal. Typically, this is achieved by keeping only a small particle of magnetized material at the apex of the tip. In practice, this condition is obtained by coating a standard tip (Fig. 11.8.4a) with the magnetic material, masking a small region of the tip apex, and then etching away the magnetic coating in the remaining portions of the tip. Such a super tip (Fig. 11.8.4b,c) will produce superior resolution MFM images (Fig. 11.8.4 d–g). For completeness, it must be mentioned that there are other interesting developments in magnetic imaging that combine the imaging power of magnetic resonance imaging (MRI) and the sensitivity of the AFM to create a hybrid

Figure 11.8.2 Schematic of one approach to implement magnetic force microscopy involving two scans. The first topography scan is carried out close to the surface in the region of strong van der Waals forces. The second scan uses the results from topography scan to keep the probe at constant height, but at sufficiently larger tip–specimen distances where only the magnetic force dominates to obtain the MFM image.

OUP CORRECTED – FINAL

784 Scanning Probe Microscopy (a)

(b)

Figure 11.8.3 Images of a magnetic array of Fe nanoelements, each of height ∼420 nm and width ∼120 nm, and fabricated by e-beam lithography. (a) An AFM image in the tapping mode, showing the excellent geometric pattern of the array. (b) A MFM image with a lift height of ∼50 nm, showing one of the possible magnetic “spin ice” configuration in this frustrated lattice system. Note that each element, shown by its outline, has a single domain configuration and is characterized by a pair of black and white “dots” corresponding to their magnetic polarity. Adapted from [44] and [47].

(a)

(d)

(e)

20 μm (b)

(c) (f)

2 μm

(g)

0.2 μm

Figure 11.8.4 Improved resolution obtained in MFM by optimizing the magnetic coating on the cantilever tip. Scanning electron micrographs of (a) a standard fully coated tip, and (b) low magnification and (c) high magnification of a supertip prepared by masking and etching the coated tip. MFM images taken of a hard magnetic disk surface with recording tracks at 2 μm periodicity with (d) the standard tip and (e) the super tip. The corresponding line scans, show the superior resolution of the super-tip (g), compared to the standard tip (f). Adapted from [24].

OUP CORRECTED – FINAL

Select Applications of Scanning Force Microscopy 785 technique that can resolve single atomic spins. Further details of this technique, known as magnetic resonance force microscopy (MRFM), can be found in [25].

11.8.3 Scanning Thermal Microscopy (SThM) As a consequence of the second law of thermodynamics, all irreversible processes involving energy interactions with the surroundings require that some of the energy be dissipated in the form of heat. However, the transport of thermal energy can be difficult to control but the scanning thermal microscope (SThM) overcomes this difficulty by controlling the flow of energy near points of contact, enabling investigation of thermal transport phenomenon at small length scales. Figure 11.8.5 shows a schematic diagram of a wire thermocouple AFM probe, which is the heart of an SThM. The probe is made of two thermocouple wires, electrochemically etched to form sharp tips, and then bonded together by capacitor discharge to form a thermocouple junction as well as an AFM tip. A reflecting metal strip is then attached across the two thermocouple wires for reflection sensing using a laser beam. Even though the smallest junction obtained (diameter ∼15 μm) is too large for topographical imaging, the junction often contains a sharp point that can effectively work as an imaging tip. Typically, the thermocouple probe is then calibrated by bringing it in contact with a copper surface maintained at uniform temperature. The application of this SThM was first demonstrated with topographical (Fig. 11.8.6a) and thermal (Fig. 11.8.6b) images of a metal-semiconductor field effect transistor (MSFET) in operation. Note that in the thermal image the source and drain are cooler than the semiconductor. This is because the metal electrodes are of lower electrical resistance. In addition, the heating is highest under the gate with the narrowest electrical channels. Finally, the drain side was observed to be Quadrant detector

Conductor 1

x,y,z Piezo

Laser Conductor 2 Laser reflector Thermocouple junction 10–20 μm diameter Specimen

Figure 11.8.5 Schematic illustration of a cantilever thermocouple probe, formed by a two-wire junction, used for scanning thermal microscopy (SThM). In ambient conditions, the heat transfer through the liquid meniscus dominates.

OUP CORRECTED – FINAL

786 Scanning Probe Microscopy Topography

1000.00 nm

Source

Drain

12.500

Adapted from [26].

1000 u

Gates

(a) Figure 11.8.6 Scanning thermal microscopy images of a nGaAs metal-semiconductor field effect transistor (MESFET). (a) Topography and (b) thermal images obtained using a wire thermocouple cantilever probe. Images obtained when the source and drain polarities are reversed (c, d), show that the hot spot is consistently on the drain side of the gate.

Thermal

25.000

Source Gate1

37.500

Gate2

μm

Drain

Drain

25.000

37.500

Gate1

μm

Gate2 Source

°C

°C 47.0

47.0

46.0

46.0 45.0

45.0 10

(c)

12.500

(b)

20

10 30

40

μm

(d)

20

30

40

μm

hotter than the source side, and this effect was reproduced when the polarity is reversed (Fig. 11.8.6c,d). Again, this is understandable as the depletion region is asymmetric and the electric channels are smaller on the drain side. This is but a simple discussion to illustrate the early development of SThM. Subsequent advances in probe fabrication, including single wire thermocouple probes, Schottky diode tips that can serve as both heaters and temperature sensors, and thin films temperature sensors deposited on commercial SiNx cantilever probes, have contributed to a steady improvement of SThM. These developments are reviewed comprehensively in [26] and are also discussed in Meyer, Hug, and Bennewitz (2004).

11.8.4 Applications of Atomic Force Microscopy in the Life Sciences Biological processes generally occur in a liquid environment and depend critically on the environment (pH and concentration of ions) and temperature. Moreover, biological building blocks (cells, membranes, molecules, etc.) demonstrate significant changes in structure and behavior when removed from the liquid environment and dried. The AFM, with its ability to image or measure specimens in physiologically relevant conditions—buffer solutions, body temperature (37◦ C),

OUP CORRECTED – FINAL

Select Applications of Scanning Force Microscopy 787 and varying pH—is ideally suited for studies in biology and the life sciences; see [27] and Morris, Kirby and Gunning (1999) for detailed discussions. These studies can be broadly classified into five areas. 11.8.4.1 Imaging of Biomolecules Of the four important classes of biomolecules—carbohydrates, lipids, nucleic acids, and proteins—proteins have been extensively studied by AFM because of their importance in disease and other biological processes. In fact, a particular area where AFM excels over TEM or other crystallographic methods is in the study of protein complexes under realistic biological conditions, i.e. in water or in buffer solution. A specific example [28, 29] that illustrates the versatility of AFM is the study of E. coli chaperonins, GroEL and GroES, and their complexes that play a critical role in protein folding processes. Combining different imaging modes, including contact and high-speed tapping modes, it has been shown that the GroES ring sometimes forms a lid for the GroEL, adding a height of ∼5 nm for GroEL/ES complex and affects the kinetics of the association or disassociation of the two entities (Fig. 11.8.7). III

(a)

(b)

(d)

(f)

I II IV

(c)

300 nm

300 nm

10 nm

5s

(e)

(g) III

I 5 nm

50 nm

10 nm

II

5s

5 nm IV

Figure 11.8.7 Images of GroEL and GroES chaperonins and their complexes. (a) TEM image of GroEL/ES complex via negative staining; nearly all GroEL show one or two GroES bound to them. (b) High resolution AFM image of the surface structure of the apical domains of the heptamer with a circular opening and seven elongated domains radiating outward. (c). Image of the GroEL/ES complex. Adapted from [28]. Association and dissociation of the GroEL/ES complex. (d) High-speed IC-AFM scans of GroEL in buffer solution without any GroES. (e) Height fluctuations of GroEL shows uniform traces. (f) Addition of GroES to GroEL in the buffer solution shows repeat variations in height. (g) The height varies between two values, differing by 3.6 ± 1 nm, consistent with the height difference between GroEL and GroEL/ES complex, indicating rapid association and disassociation. Adapted from [29].

OUP CORRECTED – FINAL

788 Scanning Probe Microscopy

Figure 11.8.8 Studies of DNA translocation and looping in real time by fast-scan AFM. Time-lapse images of an EcoP15I-DNA complex obtained at 1 frame per sec. The elapsed time is shown in each image. Translocation and formation of an extruded loop between 1 and 10 s, before release of the loop at 11s.

1s

2s

3s

4s

5s

6s

7s

8s

9s

10 s

11 s

12 s

Adapted from [30].

Another example is the study of DNA molecules, which are imaged by depositing a drop of solution on a freshly cleaved mica substrate. Note that both mica and DNA molecules are negatively charged under normal conditions and require positively charged cations (e.g. Ni2+ or Mg2+ ) to create a bridge between the substrate and the DNA molecule to immobilize the latter, and allowing for its imaging in air or in solution. Further, controlling the ionic concentration in the solution allows the DNA molecules to move in two dimensions while ensuring that they remain affixed on the surface for imaging. This has allowed observations [30] of protein–DNA interactions in real time (Fig. 11.8.8). 11.8.4.2 Imaging of Lipid Membranes It is well known (Karp, 2003) that plasma membranes, or lipid bilayers, separate the intracellular components from the extracellular environment of cells. Further, the membrane allows selective molecules to go in and out of the cell while blocking all other unwanted substances. Model phospholipid bilayers can be prepared by Langmuir–Blodgett methods, deposited on substrates, and the resultant flat surfaces and their phase separation [31] imaged [32] by AFM (Fig. 11.8.9). 11.8.4.3 Imaging of Bacterial Cells AFM images complement the widely used optical imaging studies of bacteria. While the optical methods provide statistically significant data on the morphology of cells and genes, AFM provides unique details of cellular properties using force spectroscopy. In particular, well-known bacterial such as E. coli [33], Bacillus [34], and Salmonella [35], have been studied by immobilizing them on a surface (Fig. 11.8.10). Furthermore, a subject of medical relevance is the interaction of antibiotics with bacteria; an example [36] of AFM imaging of the response of

OUP CORRECTED – FINAL

Select Applications of Scanning Force Microscopy 789 30 Height/nm

(a)

1 μm

20

(b) 6.4 nm 6.3 nm

10

6.2 nm 5.2 nm

0 0

(c)

(e)

3 2 Distance/μm

1

4

5

(d)

(f)

Figure 11.8.9 Imaging of bilipid layers. (a) Tapping mode (IC-AFM) image of multiple lipid bilayers on mica under a fluid. (b) Thickness of the bilayers range from 5.2 ± 0.1 nm adjacent to mica, to 6.2–6.4 ± 0.1 nm for the subsequent layers. (c, e) Topography, and (d, f) friction images of monolayers in air (c, d) and bilayers under water (e, f) of a mixture of saturated distearoyl-phosphatidylethanolamine (DSPE) and dioleoylphosphatidylethanolamine (DOPE). (a–b) Adapted from [31]. (c–f) Adapted from [32].

OUP CORRECTED – FINAL

790 Scanning Probe Microscopy (i)

(ii)

nm 200 nm

(iii)

nm 6000 12000 4000 8000

nm 400 nm 12000 8000 4000

4000

2000 nm

0

0

2000

4000

6000

Figure 11.8.10 Escherichia coli bacterial cells dispersed on mica and imaged by AFM in contact mode applying a force of 1nN showing (a) 3D height and (b) deflection mode images. (i). Image of bacteria in liquid cultivation medium (ii) Image of bacteria in air. (iii) High-resolution AFM images of pili-like fimbrial structures and flagella in the curli mutant MAE14 after 4 h at two different scan sizes. (i–ii) Adapted from [33]. (iii) Adapted from [35].

bacteria to antibiotic treatment, showing a significant reduction in stiffness after treatment, is illustrated in Figure 11.8.11. 11.8.4.4 Studies of Mammalian Cells In addition to imaging cells under physiological conditions, AFM can be used for nanoindentation or nanomechanical response studies of live cells. This allows differentiation between diseased and healthy cells as the mechanical stiffness of a cell decreases upon tumor invasion and metastasis. The difference in stiffness between healthy and cancerous cells measured by AFM (Fig. 11.8.12) has been proposed as a way to diagnose cancer [37]. 11.8.4.5 Biological Force Spectroscopy Last, but not least, force distance curves of AFM cantilevers are used to study the strength of specific antigen–antibody interactions. In fact, the binding between avidin and biotin is now used as the standard in such measurements because of both its strength and specificity [38]. A variant of such force spectroscopy is to measure [39, 40] unfolding dynamics of proteins (Fig. 11.8.13). The protein is adsorbed onto a surface, often by covalent attachment to gold through cysteine residues engineered at the C-terminus. When the cantilever is lowered onto the surface one of the protein molecules attaches to it, typically by nonspecific adsorption. Next, when the cantilever is raised away from the surface by a piezo

OUP CORRECTED – FINAL

Select Applications of Scanning Force Microscopy 791 20 min

170 min

(b)

20

(c) Frequency (%)

(a)

100 nm

40 min

210 min

15 10 5 0 0

500

1000 1500 2000 Elasticity (kPa)

2500

500

1000 1500 2000 Elasticity (kPa)

2500

80 min

Frequency (%)

20

260 min

500 nm

15 10 5 0 0

Figure 11.8.11 Direct AFM observation of bacterium Staphylococcus aureus cell wall digestion by the antibiotic Lysostaphin. (a) Low-resolution deflection image of a single S. aureus cell trapped in a pore of the polycarbonate membrane recorded in phosphate buffered saline (PBS) solution. (b) Imaging of lysostaphin-treated cells at high resolution. A series of deflection images recorded on top of a single cell after incubation with 16 μg/mL Lysostaphin for 20, 40, 80, 170, 210, and 260 min is shown. (c) The effect on cell wall stiffness with (top) no, and (bottom) 80 minutes of treatment. Adapted from [36].

16

Counts

(a)

(c)

12

6

0

0

9

Counts

(b)

1

2

3

4

5

6

(d)

6

3

0

0

1 2 3 4 5 6 Young’s modulus, E (kPa)

Figure 11.8.12 Can the AFM diagnose cancer cells? Optical image of (a) the cantilever and a mixture of normal and cancer cells and (b) 12 h ex vivo cultured cells demonstrating the round, spherical morphology of visually assigned tumor cells (“tumor”) and the large, flat morphology of presumed benign mesothelial (“normal”) cells; scale bar = 30 μm. Histograms of cell elastic stiffness for (c) tumor, and (d) normal reactive mesothelial cells. Adapted from [37].

OUP CORRECTED – FINAL

792 Scanning Probe Microscopy (b) 350 (a)

300

Piezo Positioner Force (pN)

LVDT

4

250 L

1

200

2 3

100

Laser

F

150

50

z

0

Split Detector

0

50

100 150 Distance (nm) t

Apply & maintain constant force, 40 pN L

50 nm

Protein

Length (nm)

(c) Solvent Droplet

200

2s

Gold coated slide Time (s)

Figure 11.8.13 Use of the AFM to study protein folding and unfolding in the (b) constant velocity, and (c) constant force modes. (a) Schematic diagram showing the operation of the AFM. (b) A “typical” AFM trace with the instrument used in constant velocity mode to determine the unfolding force at a given pulling speed. The red trace represents the cantilever deflection during the approach to the surface, and the blue trace during the retraction. A sequence of peaks is observed in the force-distance curve. The first peak (1) reflects detachment of the tip and/or protein from the surface. Force is exerted on the protein until one domain unfolds (2). The unfolded polypeptide chain is then stretched (3) until another domain unfolds. The distance between unfolding events (marked as L) is characteristic of the unfolding of a domain of ∼90 amino acids. The domains unfold at a force, F∼200 pN, as determined from the height of the unfolding peaks. Finally, the protein detaches from the cantilever (4) when all the domains have unfolded. (c) A “typical” AFM trace showing the same protein being pulled while the AFM is used in constant force mode. At the time marked by the red arrow, the force was switched from –20 pN to 40 pN to initiate the measurement. At this force, the unfolding steps (L) have a mean size of ∼20 nm (the two larger steps correspond to two domains unfolding simultaneously). In this experiment, the time between unfolding events (t) can also be determined. Adapted from [39].

OUP CORRECTED – FINAL

Select Applications of Scanning Force Microscopy 793 stage, a force is exerted on the protein. The position of the cantilever and the associated force it exerts is determined using a laser and a split photodiode detector. The linear voltage differential transformer (LVDT) monitors displacement in the z-direction.

11.8.5 Dip-Pen Nanolithography (DPN) As shown schematically in Figure 11.4.5, the narrow-gap capillary formed between the AFM tip and the specimen, when an experiment is conducted in air, condenses water from the ambient atmosphere. Moreover, this is a dynamic problem, as that water will either be transported from the substrate to the tip, or vice versa, depending on the humidity and the substrate wetting characteristics. Initially, such condensed moisture was considered a nuisance for high-resolution

(b) (a)

(c)

AFM Tip

Molecular Transport

Writing Direction

μm

0

4

0

(d)

μm

7

μm

4

(e)

Water Meniscus

Au substrate 0

μm

5

0

Figure 11.8.14 (a) Schematic representation of DPN. A water meniscus form between the AFM tip coated with 1-octadecanethiol (ODT) and the Au substrate. The size of the meniscus, which is controlled by relative humidity, affects the ODT transport rate, the effective tip-substrate contact area, and DPN resolution. Lateral force SFM images of DPN patterns written on Au substrates. (b) AFM tip, coated with ODT held in contact with the substrate for 2, 4, and 16 min (left to right) with relative humidity ∼45%, and the image recorded at a scan rate of 4 Hz. (c) Dots of 16-mercaptohexadecanoic acid created by holding the coated tip in contact with the substrate for 10, 20, and 40 s (left to right); relative humidity ∼35%. Comparing (b) and (c), we conclude that the transport properties of 16-mercaptohexadecanoicacid and of ODT differ substantially. (d) An array of ODT dots (∼20 s contact each) generated by DPN. (e) A molecule-based grid of lines 100 nm wide and 2 mm in length, written in a time of 1.5 minutes. Adapted from [41].

OUP CORRECTED – FINAL

794 Scanning Probe Microscopy AFM imaging. However, this liquid transport mechanism has been transformed into a nanolithography method [41], using the AFM tip to function as a nib (as in a fountain pen), initially coated with molecules (the ink), having a specific affinity for a substrate surface such as gold (the paper), to write patterns—using the x-y piezo drives—of these specific molecules on the sub-micron length scale. If the molecules, first coated on the tip, are chosen such that upon transport they anchor themselves by chemisorption on to the substrate surface, stable surface structures/patterns can be formed. This method, first demonstrated using 1-octadecanethiol (ODT) molecules and a gold substrate, and schematically illustrated in Figure 11.8.14a, has been termed dip-pen nanolithography (DPN) by its inventors [41]. The uniform diffusion properties of two different inks are illustrated in Figure 11.8.14, for ODT (b), and an alkanethiol derivative, 16 mercaptohexa-decanoic acid (c), proving that this is a general technique applicable to a variety of “inks”, and capable of writing different arrays of dots (d) and patterns (e). The resolution of DPN depends on the grain size of the substrate (similar to the texture of paper affecting conventional writing with a pen), and moreover, the self-assembly and chemisorption of the molecules on the surface can be used to control “ink” diffusion. Over the years, DPN patterns have been created with a number of biological molecules including proteins, peptides, and DNA, with applications of such arrays in proteomic and genomic testing [42].

Summary SPM involves scanning a fine probe/tip very close to a specimen surface and measuring either the quantum-mechanical tunneling current (STM) or a force (SFM) based on one of many possible tip–surface interactions. In SPM, both the tunneling current and the force can be measured at high spatial resolution, as a function of tip position, using piezoelectric actuators to move the tip in three orthogonal directions. The tunneling current across the vacuum gap between the tip and surface provides either an image at atomic resolution (microscopy) or details of the local electronic structure of the surface (spectroscopy) measured as a function of specimen bias voltage. The tunneling tip can also be used to manipulate adsorbed atoms on a clean surface. The STM can be operated either in a constant current mode using a feedback loop to the piezo stage to control the tip height, or in a constant height mode (requires atomic flat specimens) and measuring the variable tunneling current. In the SFM, a force sensor—a sharp tip mounted on a vibrating cantilever— replaces the tunneling tip of the STM. Short range (100 keV)

•

Lateral resolution ∼200 μm × 4 mm

Accuracy: Lattice parameters to four significant figures using CBED

Space: typically attached to a UHV thin film growth chamber

Space: 3 m × 3 m

XRD §7

X-RAY DIFFRACTION Crystallography, including atomic arrangements in the unit cell. Identification of crystalline phases. Strain, orientation and crystallite size determination

• •

X-rays in, X-rays out Analyze all elements, but low Z elements (particularly H) difficult. Not element specific

• •

Monolayer sensitivity with synchrotron radiation

• •

Nondestructive for most materials

•

Satisfies Bragg’s law (real space) or the Laue criterion (reciprocal space)

Detection limit ∼3% in two-phase mixtures

•

Lateral resolution is poor, but with microfocus ∼10 μm

Depth ∼few μm, but depends on the material

• •

Any material Analysis can be done in air

• •

Cost: $50K–$400K Space: ∼8 m2

OUP CORRECTED – FINAL

817

818 OUP CORRECTED – FINAL

Table 12.3 Imaging Methods. Acronym and Section

METHOD and Use

Salient Details

Sensitivity, Limits, Resolution, etc.

Specimen

Cost and Space Requirements

APT §1.4.3

ATOM PROBE TOMOGRAPHY Three-dimensional imaging of a fine tip specimen with sub-nm resolution

•

Destructive, removes atoms by successive evaporation

•

Best resolution ∼0.2 nm, typical 0.3 nm

•

Cost: ∼$0.5M–$1M

•

Evaporation with either voltage or laser (Nd: YAG, 1064 nm; or Nd:YVO4 , 355 nm) pulses

•

Depth inversely related to laser wavelength

Sharp, needle-shaped solid specimen (tip radius ≤100 nm)

•

Best prepared by FIB methods

CONFOCAL SCANNING OPTICAL MICROSCOPY Optical imaging using a spatial pinhole to block out-of-focus light in image formation

• • •

Photons in, photons out

Resolution ∼0.2 μm (x, y), 0.5 μm (vertical)

All specimens, including biological ones

ELECTRON HOLOGRAPHY IN A TEM Quantitative evaluation of the magnetic induction or electric potential of a thin foil specimen at high resolution

• •

Electrons in, electrons out

•

Specimens should be electron transparent and vacuum compatible

CSOM §6.8.5

Electron Holography §9.3.7

• •

Operate in reflection mode Scan either the specimen or the beam

Spatial resolution ∼20 nm (best ∼5 nm)

Preparation of electron transparent specimen may damage microstructural features

•

Thickness integrated (≤100 nm) signal

Requires HV condition.

•

Image acquisition time ∼50 msec–10 sec

Requires a coherent source (FEG or SEG)

• •

Cost: ∼$1M–$2M Space: requires vibration isolation area ∼3 m × 4 m

Electron Tomography §9.5.1

HREM §9.3.5

ELECTRON TOMOGRAPHY IN A TEM Technique for reconstructing three-dimensional images from a series of two-dimensional projections (images)

HIGH-RESOLUTION ELECTRON MICROSCOPY Phase contrast images of thin foils formed by interference of the direct transmitted beam with one or more diffracted beams, selected by an aperture

• •

Electrons in, electrons out

•

Requires true projection, i.e. intensity must show monotonic relationship to some function of thickness or density.

•

Particularly suited for studying amorphous, polymer, and biological materials

• •

Electrons in, electrons out

•

Often requires simulations for interpretation

•

Resolution defined in terms of the contrast transfer function

•

Best transfer function, with minimum number of zero crossings, occurs at the Scherzer defocus

•

Correction for spherical and chromatic aberration possible

Consists of acquisition of the projection series, tomographic image reconstruction, and final image viewing and segmentation

• •

Resolution ∼10 nm

•

Resolution ∝ CS 1/4 λ3/4 , improves with smaller λ (higher keV) and smaller CS

•

Typically, resolution ∼0.1 nm

Image is a function of both defocus and thickness

Requires large specimen tilt range single axis ∼ ±70◦ , dual axis ∼ ±50◦

Electron transparent specimens of inorganic materials and biological specimens.

• •

Cost: ∼$1M–$2M Space: requires vibration isolation area ∼3 m × 4 m

Thin specimen ≤20 nm

continued

OUP CORRECTED – FINAL

819

820 OUP CORRECTED – FINAL

Table 12.3 Continued Acronym and Section

METHOD and Use

Salient Details

Sensitivity, Limits, Resolution, etc.

Specimen

Lorentz Microscopy §9.3.6

LORENTZ MICROSCOPY IN A TEM Imaging magnetic domains in a TEM

• •

Electrons in, electrons out

•

Three possible methods: Fresnel and Foucault (F&F), and Differential Phase Contrast (DPC)

Spatial resolution ∼10–50 nm (F&F), ∼2–10 nm (DPC)

Electron-transparent specimen.

• •

Thickness integrated, t ≤ 100 nm.

OM §6.8

OPTICAL OR LIGHT MICROSCOPY Method of choice for the first visual observation of a specimen and used for metallography

•

Qualitative (F&F) and quantitative (DPC) measurement of in-plane induction

• • •

Photons in, photons out

•

Different imaging modes include using cross-polarized light, differential interference contrast, dark field, and phase contrast

Nondestructive Reflection and transmission geometry

Cost and Space Requirements

Image acquisition time ∼50 msec–60 sec.

Resolving power ∼0.20 μm using white light

•

All solid, liquid, and biological specimens

•

Image quality in metallography enhanced by polishing/etching surface

•

Cost: ∼$2.5K–$50K

•

Space: ∼ Tabletop

SEM §10

SEMPA §10.5.2

SCANNING ELECTRON MICROSCOPY High-magnification imaging and chemical analysis, including mapping

SCANNING ELECTRON MICROSCOPY WITH POLARIZATION ANALYSIS. Maps local magnetization and domain structure of surfaces

•

Electrons in, electrons out (imaging)

•

Magnification range ∼10×–300,000×

•

Electrons in, photons out (chemical analysis)

•

Lateral resolution ∼1–50 nm for secondary electrons

•

Nondestructive, but specimens may be subject to beam damage

•

•

Incident beam energy ∼0.5 keV–30 keV

Depth sampled varies from nm to μm, depending on kV, sample and analysis mode

•

Large depth of field, 10–10,000 μm

• •

Electrons in, electrons out

•

Spatial resolution ∼150 nm (best ∼20 nm)

• •

Depth of information ∼1–2 nm

• • •

Typically uses a Mott analyzer to detect secondary electron polarization Best to have electrostatic lenses

Time for image acquisition ∼1–100 min

•

Specimens may need a conducting coating to avoid charging effects

•

Size 0.1 mm– ∼10 cm

•

Specimens should be vacuum compatible

•

A clean surface is required

•

Cost: $100K–$300K

•

Space: ∼2 m × 3m

•

Cost: $300K–$500K

•

Space: ∼2 m × 3 m.

Requires UHV environment Quantitative evaluation of magnetization distribution continued

OUP CORRECTED – FINAL

821

822 OUP CORRECTED – FINAL

Table 12.3 Continued Acronym and Section

METHOD and Use

Salient Details

Sensitivity, Limits, Resolution, etc.

Specimen

Cost and Space Requirements

SFM §11.4

SCANNING FORCE MICROSCOPY Imaging in air, vacuum, or solution, and measuring surface forces with excellent resolution and accuracy

• • •

• •

Vertical resolution ∼0.01 nm

•

Conducting or insulating

•

Lateral resolution ∼ atomic–0.1 nm

•

•

Typical force measurement ∼1 nN

Biological materials

Cost: $75K (air)–$200K (UHV)

•

•

Specimen can be in liquid or in air

•

Vibration-free environment required

Space: Instrument ∼0.2 m × 0.2 m sits in 0.6 m × 0.6 m × 0.6 m vibration isolator

•

Flat, conducting specimen.

•

•

Insulators can be viewed with conducting coating

Cost: ∼$150K including vacuum system

•

Space: ∼ table-top

•

Vibration-free environment

STM §11.2

SCANNING TUNNELING MICROSCOPY Imaging surface topography at atomic resolution, and measuring the local electronic structure in spectroscopy mode

Nondestructive Cantilever used as force sensor Probe physically interacts with the specimen

•

Nothing to focus and zero depth of field!

•

•

Imaging in ambient and liquid environments

Maximum scan area 100 × 100 μm

•

Time for image ∼ 2–5 min

• •

Various forces can be measured

• •

Nondestructive

• •

Vertical resolution ∼0.001 nm

• •

Can determine work function

•

Requires high vacuum

Contact, noncontact, and tapping modes

Tunneling current probes both occupied and unoccupied density of states depending on bias Contaminated tips can create artefacts

Lateral resolution ∼ atomic scale

TEM §9

XTM §6.6.7

• •

Electrons in, electrons out Elastic scattering (imaging and diffraction) and inelastic scattering (EDXS and EELS)

•

Magnetic contrast (Lorentz microscopy and electron holography) possible

•

Reciprocity theorem relates standard TEM and STEM modes

•

Characteristics depend on source/gun parameters

•

EM lenses corrected for aberrations in high-end machines

X-RAY TRANSMISSION MICROSCOPY Generate microscopic images of a thin specimen by scanning it in a focused X-ray beam

• •

Photons in, photons out

•

Use dichroism for magnetic contrast

•

Element-specificity in imaging

Z-CONTRAST IMAGING IN A STEM Image formed by large angle Rutherford scattering with element sensitivity (∝ Z2 )

• •

High-angle incoherent scattering

•

Scattering angles ∼ 70–150 mrads

• •

Resolution ∼0.3–5 Å

•

Convergence semi-angle, α ∼ 10–2 rad, scattering semi-angle, β ∼ 15 mrad

Thin specimen (≤100 nm)

•

Preparing electron transparent foils is a challenge

Resolution using zone plates ∼10 nm

Thin specimens, t ≤ 20 nm

Can resolve single atomic columns using sub-Å probes and aberration-corrected lenses

Thin electron-transparent specimens

Use zone plates for X-ray focusing

Combines a STEM with a high-angle annular dark field detector to collect electrons incoherently scattered over large angles

•

Cost: ∼$0.5M–$2M and more!

• •

Space: requires vibration isolation area ∼3 m × 4 m

OUP CORRECTED – FINAL

Z-Contrast §9.3.4

TRANSMISSION ELECTRON MICROSCOPY Observing the structural/crystallographic, chemical and magnetic microstructure of a specimen, with diffraction (SAD, CBED), imaging (mass, diffraction, and phase contrast), and spectroscopy (EELS, EDXS) at the highest possible lateral resolution

823

OUP CORRECTED – FINAL

OUP CORRECTED – FINAL

Index

( fn denotes foot-note)

A A, Z, F correction factors 117–18 Abbe, Ernst 376fn aberration-corrected microscope (TEAM) 45 STEM instrument 562, 600 aberrations (lens) 378–9 chromatic aberration 378, 550, 559, 563 spherical aberration 541, 559, 562, 581 abrasive materials 3f absorbance A, Beer–Lambert law 163 absorption cross-sections 199 effect A 117 mass absorption coefficients and densities, select elements and X-ray radiations 102t, 104f mean escape depth of photoelectrons 199 absorption spectroscopies, probing unoccupied states 197–9 absorption spectrum, Mo-K edge 203–4 absorption or transmission factor 453 accelerating electric charge 58 acceleration voltage 718f accelerators 291–2 activators, promoting cathodoluminescence 732 adsorption on surfaces 487 adsorption on surfaces, using LEED 493–6 advanced light source (ALS) 106fn, 107f

Advanced Photon Source (APS) 105, 106fn, 277f Ag polycrystalline Ag UPS spectrum 195f as target material, X-ray tube 100 air pollution, PIXE spectrum 336f aircraft structures, polymer-matrix composites, FTIR 179 Airy disc 365–6, 376 Airy, George 365fn Al back-scattering yield 319 sputter yield data for aluminum 313f Al2 O3 LEISS and RBS spectra 325f ruby solid-state laser 281f simulated HREM images 581f albite 16 algebraic operations on images 53 alloys dual-phase steel, microstructure 20–1f electrochemical polishing 666 intermetallic alloys using EFTEM images 654f long-range order 464 chemically ordered examples 464 unit cells of ordered binary alloys 465f alumina see Al 2 O3

American Society of Testing and Materials (ASTM International) 7fn amphiboles 16 amplitude ratio in ellipsometry 399 amplitude-modulated atomic force microscopy (AM-AFM) 774, 775f atomic resolution imaging 781–2 block copolymer mesophase 775f amplitude-phase diagram 511f analytical electron microscope (AEM) 556–7f, 617–19 electron energy-loss spectroscopy 623–41 energy dispersive X-ray microanalysis 641–8 energy filtered imaging 637–41 summary 556 analyzer, used for optical imaging 384–5f angle of detection of back scattered electrons, defined 715 angle-resolved photoemission spectroscopy (ARPES) 155, 188 angular component of probability distribution of electron orbitals 75f angular dependence, back-scattered electron density distribution, topographic information 715–16

angular distributions of electron scattering in transmission 625f angular momentum 69 quantum number 85 spin angular momentum 72 anisotropy 389fn, 397 annealing temperature 21f annular dark field (ADF) detector 585, 599 image 586 annular detector 716f anorthite 16 anthracene, UV-Vis 165f antibiotics, Staphylococcus aureus and 791f antigen–antibody interactions 790–1 antiphase boundary (APB) 249 applied voltage, hysteresis and creep of piezoelectric drives 776 4 Ar+ ions, sensitivity to neutralization 326 Argand JR 400fn Argand diagram 351 aromatic ring, Raman frequencies 183t astigmatism defined 308 images 606–7 asymmetric object 240 atom, energy levels for multi-electron atom 87 atom probe field ion microscope (APFIM) 46f atom probe tomography (APT), summary 818t atomic displacements on electron impact 298–9

OUP CORRECTED – FINAL

826 Index atomic electron binding energies 35 atomic fingerprinting 745f, 748 frequency modulated atomic force microscopy 781–2f atomic force microscopy (AFM) 5t, 47–8, 745f, 747 amplitude-modulated (AM-AFM) 773–6, 781–2f atomic fingerprinting 782f biological sciences 786–93 cantilever as force sensor 760f, 763–5, 790–1 dynamic noncontact modes 774 first 760f frequency-modulated (FM-AFM) 774, 774f, 781–2f in operation, principal components 777f piezoelectric scanners 778f tapping model, intermittent-contact AFM (IC-AFM) 759, 774–5f, 789 atomic and mass fractions 727 atomic mass unit 825t atomic number effect (Z) 116 in X-ray microanalysis (Z) 115 average on secondary electron yield 714 binding energy, all elements 131f electron diffusion depth 708f Monte Carlo simulations of electron propagation 592 Z-dependence of back scattered electrons 720 atomic orbitals 149–50f linear combination of, (LCAO) 149 atomic resolution imaging force-displacement curves for Si Sn Pb 781–2 Z-contrast, EELS and X-ray spectra maps 601f

atomic resolution microscope (ARM) 45f atomic scattering factor 414–18, 452, 482–5, 505, 591 Cromer–Mann parameters 421t defined for electrons 482–5 for neutrons 535–8, 536t for X-rays 419t atomic spectra, transitions emissions, and secondary processes 78–97 atomic structure 69–78 electronic binding energies 79t electronic structure 74t hybridization 77f spin-orbit interaction 77 atoms absorption and emission 34–6 interactions 35f interatomic bonding in solids 152–7 manipulation of adsorbed atoms on clean surfaces 757–9f multi-electron atom 35f transitions, nomenclature 85f Au back-scattering yield for 1.4 MeV 4 He+ ions 321 FluoroGold 655fn islands on a C film 586f Auger, Pierre 91 Auger decay 78 Auger electron emission Coster–Kronig process 92 kinetic energy 92–3 KLL lines 94t non-radiative 36–7f, 91 Auger electron spectroscopy (AES) 31f, 78, 123–4, 140 comparison of AES and XPS 136–7 compound AES and XPS system 121f

energies of principal Auger lines 94t, 124f Omicron-HAXPES system 122f sensitivity factors for quantitative analysis for 5 keV electron probes 135f spectrum of Co surface probed with 2 keV electrons 126f sputter depth profiling of Ta–Si film 128f summary 804t Ti–Al superalloy specimen mapping 128f and XPS system 121f Auger lines 94t, 124f automobile exhaust, Pt-based three-way catalyst analysis 208 Avogadro number 148, 150, 590, 825t axial shift defining depth of field 704

B Ba, mass absorption coefficient 104f back focal plane (BFP) 584–5f back-scattered electrons (BSE) channeling 720f density distribution, topographic information 715–16 diffraction 720–1 orientation of grains in polycrystalline specimen 717f yield 716f back-scattering images surface-tilt and shadow contrats 714f probability channeling 721 SEM 296, 727 spectroscopy, energy width 318 yield as a function of Z 712f bacterial cells, atomic force microscopy (AFM) 787f, 788–790

band energies 150 valence s and p electrons hybridization 76f band gaps 700 semi-conductors 152f band structure 155 band-pass analyser 120 barns, units cross-sections 305f BaTiO3 wavelength and energy dispersive X-ray spectra 109f Be window 111 beach marks, failure of engineering materials 715f beam incident beam energy 708f depth resolution 32 point incident beam, trajectory envelopes 708f see also electron beam beam broadening, elastic and inelastic scattering 707f, 708f beam convergence semi-angle compared to Bragg angle 524f effect on transmission electron diffraction 524f beam deflection coils, Lorentz force 697f beam electrons 619fn generalized oscillator strength (GOS) 628 inner-shell ionization 620 beam heating 301 beam–solid interactions electron 588–617 Elastic interactions 589–92 Mass-thicknes contrast 592–5 Diffraction contrast 595–8 Z-contrast 598–601 Phase contrast 601–9 Magnetic contrast 6104 signals, TEM and SEM 699f –700 Beer–Lambert law 163, 164, 166 bend contours 511–12f

OUP CORRECTED – FINAL

Index 827 benzene naphthalene, anthracene, UV-Vis 165 photoemission spectra 192 Bessel functions 365 Bessemer process vs open-hearth process 14–15 Bethe, HA 503fn Bethe ridge 628 bias voltage tunneling 751 binding energies electrons atomic number effect (Z) 131f electrons in atoms 79t elements 79t, 87f, 131f energy level diagram, Cu 88f chemical shifts observed in XPS 132f Na and Mo 86f oxidation states 194 photoemission spectroscopy (PES) 195 two-electron (ABC) de-excitation processes 92f Binnig, Gert 47fn, 746fn, 763 biological force spectroscopy 790–3 biological tissues computed axial tomography (CAT) 650–5 specimen preparation 667–70f see also proteins biomolecules, AFM 786–93 bisphenol polymer, applications 179 body centered cubic (BCC) structure 20–1f, 233–5f, 427, 428t single crystal spot transmission electron diffraction patterns 517f Bohr, Niels HD 69fn Bohr model, atom 34 Bohr radius 23, 70, 591, 620, 825t for H 825t Boltzmann constant 283 bonds/bonding 147–214

bond strength 158 bond types 148f covalent bonding in solids 152–7 equilibrium bond energy 158 interatomic bonding in solids 148, 152–7 ionic bonding in solids 153–4 metals 156–7 wave numbers, bending and stretching vibrations 172t Born approximation 483 Born, Max 483fn bound water index (BWI) 539 Bragg, Sir Lawrence 39fn Bragg diffracted beam 578 Bragg reflection 433–4f, 656–8 Bragg scattering angle 259f, 434–5, 437, 442, 455, 457f, 460, 503, 507, 509, 524f, 569, 649, 656, 721 high energy electrons 577–8, 592 vs beam convergence semi-angle 524f Bragg’s law of diffraction 39, 108, 256–62f, 422, 434–5 Debye–Scherrer method 438 diffraction spots 520 Ewald and limiting sphere construction 259–60f, 489 lattice planes 424f Laue method 438–9f path difference 256 SEM 721 X-ray and electron diffraction, comparison 262–266 Bragg–Bretano geometry thin films and multilayers, diffractometry, reflectivity, and pole figures 445–9 X-ray diffraction 440f Bravais, R 224fn

Bravais lattices 224–6f, 225t, 464 symmetry 236 breaking radiation 97, 101, 450 bremsstrahlung 97, 101, 449–50 Brewster angle 396, 407 bright field (BF) imaging 512f, 572f –3, 575f, 585 specimen preparation 666 electron detector 584 grain boundary phases 19, 576f two-beam condition of edge dislocations 658f brightness, defined 285 bulk analytical tools 89 Burgers vector 247–8, 596, 656, 657

C C=O mixed with NH, deformation Raman frquencies 183t Caenorhabdites elegans, electron tomography 654–6 CaF2 (insulator) 176f calibration plot Si-Ge Raman spectroscopy 184f cantilever AFM force–distance curves 790–1 deflection 769 force balance affecting displacement 770f mechanical characteristics 760–3 cantilever probes 759–2f, 761fn function as deflection sensor 762, 762f, 769 function as force sensor 763–5 thermocouple probe, SThM 785f capillary force, scanning force microscopy 767 carbon dioxide stretching modes 171 variation of dipole moment and polarizability 171f vibration modes 162f

carbon film, diffractograms under different imaging conditions 607f carbon (forms), X-ray absorption spectrum 218 catalyst, three-way (TWC) 208 catalytic particles 56f analysis 208–9 cathodoluminescence (CL) 16 high-energy 700f in semiconductors 731–2 summary 804t Ca–Si bonding, PES 196 cement bound water index (BWI) 539 hydration, time-resolved neutron-scattering 538–40 CEOS aberration correctors 587f ceramics grain boundary phases, bright vs dark field imaging 19, 576f high resolution atomic structure images 45f MSE 3f piezoelectric lead– zirconium–titanate, lead–barium– titanate 776 sintering and grain boundary phases 17–19 cesium chloride, structure 429–30 CH3 umbrella mode, Raman frquencies 183t channeling effects, lattice anisotropy 708f chaperonins, GroEL and GroES 787 Charpy impact fracture surface 14f characterization and analysis examples 7–21 methods: spectroscopy, diffraction, imaging 34–48 probe based methods 5t probes and signals 22–34

OUP CORRECTED – FINAL

828 Index chemical order in materials 529–30 chemical order–disorder transitions 464–6 chemical shifts 629 chemical X-ray microanalysis 113, 642 SEM 726–30 ZAF corrections 116 k-factors 118, TEM 641–9 chemical/electrochemical polishing/etching 668 semiconductors 666 chromatic aberration 559 chromatic aberration coefficient 378 chromophores 163fn cleavage fracture, failure of engineering materials 715f Cliff–Lorimer k-factors 643 CMOS technology 664 coherence length/area 496 coherent, defined 58 coherent elastic scattering 595 incoherent and inelastic scattering 504–5 cold field emission tip 283f parameters 284t collimators and monochromators 449–50 colloids, groundwater 138–9 column approximation dynamical diffraction 509–14, 510f kinematic scattering 541 common functional groups, Raman frequencies 183t common line, zone axis 232f compounds, positive ion yield as a function of atomic number 314f Compton, AH 410fn Compton incoherent scattering 413fn–14 computed axial tomography (CAT) 650 weighted back-projection method 651

computed tomography, X-ray attenuation 16 concentric hemisphere analyzer (CHA) 122 condenser lens effects on electron beam 695 design for thermionic and field emission guns 694 two-condenser lens system 696f conduction electron bands 150 conduction electron (valence) bands, plasmons 184, 620–1f confocal scanning optical microscopy (CSOM) 44, 386 summary 818t conjugated molecules 163 constant energy photon probe 188 contrast transfer function (CTF) 579–80f, 582 phase contrast imaging 575 conventional TEM (CTEM) vs STEM 584–5f convergent beam electron diffraction (CBED) 482, 523f –6, 541, 557f, 644 applications 649–50 radii of higher-order Laue zones (HOLZ) 649 convolution STM 754f core electron, basic edge shapes 629 core hole 35f Coster–Kronig process 92f coulomb interaction 304, 593, 598 fast electron beam with +Ze of atomic nucleus 598 coulomb-volt 23 coulombic attractive force 69, 588 coupling, spin–orbit interaction 76, 77f covalent bonding in solids 154–5f Co–silicide 664f Crewe, Albert 584fn Crick, Francis 12fn

Cromer–Mann parameters 421t, 426, 427 cross-section calculated for different sub-shells 136f differential scattering cross-section 591 ions 304 elastic-to-inelastic scattering ratio 622f –3 ionization Q 621 plasmon excitation 620 Rutherford back-scattering 598 scattering 589–91, 620–1 mean free path length 589–90 specimens 669–70, 736 crossed polarizers, for non-cubic materials 384–6 crystallite shape effect on shape transform and diffracted intensities in transmission, for thin plates 510f –14 or grain size and lattice strain measurements 460 shape transforms 509 crystallography 57, 220–56 stereographic projection 244–5f symmetry and International Tables for Crystallography 239–44 crystals/crystalline state 7, 222–3 10 point groups in two dimensions 241f 17 space groups in two dimensions 242f 7 crystal structures 226f 14 Bravais lattices 226f factor for X-ray scattering 422 atomic size, coordination,and close packing 233–6 body centered cubic (BCC) structure 233–5f, 427 defined 221, 268, 482

new definition 267 defocused computer simulated images 581f diffraction of polycrystalline film 446–7 diffraction of single crystal film 447 dislocations and stacking faults 604, 655–6f, 658t, 664 energy bands in a periodic potential 283f epitaxial growth 195fn examples of srystal structures 236–9 face centered cubic (FCC) structure 20–1f, 233–5f, 428t, 523f, 548f generalized systems and Bravais lattices 224–5t growth mode, thin films 498 hexagonal close-packed plane of atoms 234f hexagonal crystals system 230f, 254 imperfections in crystals 247–50, 509–14 indices for directions 228f interplanar spacing, all systems 230t ion channeling 324f lattice points, lines, directions, and planes 227–9 lattices 223–4, 250–2 nanoparticles 8 orthorhombic system 243f, 254 periodic arrangement of scattering sites 39f pole figures for MgO films 447 polycrystalline material 222f repeat distance, period d 38 single diffractometer settings 446f diffractometry 438–9 structure analysis 20 summary 267–8

OUP CORRECTED – FINAL

Index 829 systems, symmetries and point groups in three dimensions 225t zonal equations 231–3, 517f CsCl structure 236–7 Cu energy level diagram 88f intensity distribution of secondary electrons 711f shake-up peaks and multiplet splitting 134f structure 236–7 white lines 200 Cu alloys BCC and FCC 464 Cu–Al eutectic SEM images at different magnifications 718f Cu dihalides, XPS spectra 193, 194f Cu K-alpha radiation, high angle scan of multilayer structure 446 Cu X-ray tube Ni filter thickness required 104t, 105 schema of X-ray radiation with and without filter 104f Cu–Zn specimen, SEM compositional maps 46 cubic structures (FCC and BCC) 20–1f, 233–5f, 427, 428t cylindrical mirror analyzer (CMA) 122

D dark field (DF) imaging 574–5f, 584 annular dark field (ADF) detector 585 BF/DF imaging, specimen preparation 666 grain boundary phases 19, 576f Davisson, CJ 26fn de Broglie, Louis 26fn, 69fn de Broglie relation 26 wave—particle duality 26, 69

wavelength 265, 303 de-excitation processes 85–97, 641 Debye, PJW 452fn Debye–Scherrer camera/method 453–5f, 458t corrected intensities 457–8 powders 443–5f, 458t ring diffraction pattern 515f, 602 Debye–Waller factors 452–3t thermal parameters 464 defect state and cathodoluminescence 700 defective crystals, dislocations and stacking faults 604, 655–6f, 658t, 664 defocus images 577f –80, 602, 605f Scherzer defocus and resolution 580–4, 605 demagnified image of source in a TEM 566 density of states (DOS) 150–1f, 756f band gap 151f, 152f, 154f, 700f N(E)in the free electron model 151 dental and medical specimens, X-ray fluorescence spectroscopy (XRF) 137–8 depth of field, secondary electron imaging 719f detectors, germanium detectors 112f deviation parameter 507f, 507–9, 549 diamond artificial, secondary electron image 721f bonding in 77f displacement energy 299t electron-yield in C KVV Auger electron spectra 127 ion channeling 324f structure 238, 244 X-ray absorption spectrum 218

diamond cubic crystal structure 518–19 single crystal spot transmission electron diffraction patterns 517f, 518f diatomic molecule, vibrational energy levels 158f, 159 dichroic mirror 387fn dielectric behavior, plasmons 620 dielectric constant 393 differential conductivity in scanning tunneling spectroscopy 757 diffracted beams 432 broadening 433–7 diffracting crystals, wavelength dispersion spectrometer (WDS) 109t diffraction beam broadening 433–7 coherent scattering 592–3 Debye-Scherrer method 553–5 long-period multilayers 445–9, 531–2 powders and single crystals 439–43 thickness fringes 511, 524, 526f see also area —; convergent beam-energy electron —; low-energy electron —; selected area —; X-ray — diffraction angle 496 diffraction contrast 595–9 excitation error 507, 596 parallel incident illumination 597 in STEM vs TEM images 598f diffraction coupling in EFTEM 632 diffraction grating 38f, 173f, 363–70 2-slit interference 38f, 366–8 periodic grooves 173 Raman scattering 181

resolving power 369–70 diffraction intensity, factors affecting X-ray 451–9 diffraction and scattering methods 814t–17t summary 814t diffractometry corrected intensities 457–8 intensities, determining structure 431 of powders and single crystals 438–41 reflectivity and pole figures 445–9 diffuse scattering, short-range order and 467–8f diffusion depth, atomic number Z 708–9 digital cameras 514fn digital imaging 48–56 acquisition 48f –50 digitization and storage 50 high-pass/low-pass filters 53, 55f image file size IFS 50 image segmentation 55 pre-processing, look-up tables, histogram equalization point, and kernel operations 49–53 processing, analysis, and output 48 resolution and quantization 50f segmentation step 49, 55–6 digitization, digital imaging 48, 50 dimensionality 2 dimple grinding, ion milling 667f dioleoylphosphatidylethanolamine (DOPE) 789f diopside 16 dip-pen nanolithography (DPN) 767, 793–4 dipole moment 167f changes 170 induced 167 variation and polarizability 171

OUP CORRECTED – FINAL

830 Index dipole selection rules 78–80, 88f, 91, 100 electron transitions 78–80, 85, 88f, 100 X-ray photons 91 Dirac, P 70fn dislocations 604 edge two-beam condition, bright field imaging 658f loops 299 perfect,in FCC crystals 658t and stacking faults 604, 655–6f, 658t, 664 dispersion of electron spectrometer 631 displacement energy 299 dissociation energy 159 distearoylphosphatidylethanolamine (DSPE) 789f DNA structure 10–12f atomic force microscopy (AFM) 788 double helix 401f protein—DNA complexes 788 X-ray diffraction (XRD) photograph 11f, 408f domains, magnetic 4, 611f, 722f double diffraction 541, 608 dual column FIB-SEM 661–2f dual function probes in imaging 655–6f ductile fracture, failure of engineering materials 715f dynamic measurements in situ TEM 665 in situ X-ray diffraction 468 in situ neutron scattering 538 materials characterization 57 ultrafast TEM 666 dynamic noncontact modes, atomic force microscopy 773–6 electron diffraction, column approximation 509–14

dynamical theory, electron diffraction 511

E Earth, minerals of crust 16 EFTEM see energy-filtered transmission electron microscope (EFTEM) EierlegendeWollmilchsau 8fn eigenvalues, eigenfunctions 72fn Einstein, Albert 25fn, 653f elastic interactions 589 elastic scattering 166 phase difference 506 SAD, HREM and CBED 557 elastic/inelastic scattering 294–6f, 504–5 beam broadening 707 cross-section 622f, 623 interactions, probe and signal 22f electric vector E,elliptically polarized light 29f electrochemical etching/polishing 666, 668 metals and alloys 666–7f electrochemical STM (ECSTM) 748f electromagnetic lenses 560f –5 astigmatism 563 corrections of aberrations 587 chromatic aberration 563 diffraction-limited resolution 562 Gaussian image plane 563 optimal divergence angle 562 spherical aberration 541, 562, 581 electromagnetic radiation absorption or transmittance in IR 169 E and H fields, propagation of 27 overview 58 scattering and 58 electromagnetic spectrum 293 IR 172

photons 25 probes, attributes 22–3f UV and EUV 22, 23f wavelength 23f, 172 waves,nature and propagation 27 electromigration 312 electron back-scattered diffraction patterns (EBSD) 721 summary 814t electron beam atomic displacements 298–300 inelastic interactions 617–18f interactions with specimen 694, 699f magnetic deflection coils 696 aperture, convergence semiangle 703f trajectory envelopes in specimens 708f see also beam electron beam-induced current (EBIC) 738 electron binding energies elements, of the 79t, 87f, 131f energy level diagram and allowed transitions, Cu 88f Na and Mo, energy levels 86f two-electron (ABC) de-excitation processes 92f electron biprism 556, 556–7f, 613, 614f electron channeling 720f –1 electron charge 825t electron diffraction 26, 482 atomic scattering factor 482–5 basics in a TEM 505–9 column approximation 509–510f differential scattering cross-section 591 dynamical theory 511 Ewald sphere construction, two dimensions 487–9 from surfaces 485–7

kinematical theory 506–9 patterns orientation relationship 535, 549 summary 540 single crystal in transmission 517f electron emission back-scattered electrons, variation 712f number effect, secondaries 714 photoemission 97 secondary 710–12 zero current flow 710f electron energy spectrometers and analyzers 120–1 electron energy-level diagrams, nomenclature 85, 86f electron energy-loss spectroscopy (EELS) 22, 35, 200, 556, 623–41 3D data set cube, imaging 639f angular distribution of electrons 625f background modeling 633, 633f core loss region 627 direct electron detection cameras 631 electron diffraction patterns 638f energy filtered imaging 637–41 experimental arrangement 619f fast electrons as incident probe 22 GdBa2 Cu3 O7 624f high-resolution, O K edge of bulk amorphous silicon 663f ideal energy-loss spectrum 633f inelastic interactions 617–18f inner-shell ionization edges 632–3 ionization edge threshold 624f, 629 low-loss region 624

OUP CORRECTED – FINAL

Index 831 microanalysis, inner-shell ionization edges 632–3 no-loss region 624 nomenclature 629 photodiode arrays (PDAs) 631–2 plasmon loss 624f, 626–7 quadrupole lens array 632f, 639f signal detection 630–2 spectrometer layout 632f three basic edge shapes 624f, 629 zero-loss peak 624 see also electron energy-loss spectroscopy (EELS) microanalysis; energy dispersive X-ray spectrometry (EDXS); quantitative microanalysis electron energy-loss spectroscopy (EELS) microanalysis core loss region 627–8 minimum detectable mass (MDM) 636 partial ionization cross-sections 635 using core edges 627–9 using inner-shell ionization edges 632 electron gun 282–6f, 490, 553, 558–9 brightness, defined as current density per unit solid angle 285 electron holography 556, 556–7f, 613–17 Fourier transform 615–16f magnetic materials 614–17 off-axis 616f summary 818t electron incidence cathodoluminescence in semiconductors 700f electron probe microanalysis (EPMA) 114 electron microscope, analytical (AEM) 556–7f, 617–19 lens aberration correction 587

electron microscopy 5t, 295, 557f, 699 accelerated electrons 288 analytical 617–50 applications, transmission 650–66 atomic displacements 298–9 beam heating 301 electrostatic charging 302–3 high voltage (HVEMs) 33 high-resolution 19 hydrocarbon contamination 303 irradiation damage, elastic/inelastic scattering 298f operating range, voltage 23 scanning 709–19 scanning transmission 584–7 reciprocity principle 586f specimen prepration, transmission 666–71 summary, transmission 671–4 summary, scanning 737–8 surface sputtering 300–1 see also analytical electron microscope (AEM); scanning electron microscope (SEM); transmission electron microscope (TEM) electron permittivity of free space 825t electron photoemission 97 energy levels of specimen and spectrometer 98f electron probe microanalysis (EPMA) 16, 78, 89, 114–19, 618 commercial EPMA CAMECASXFive 116f electron incidence 114 summary 806t electron probes broadening, inelastic scattering 296f mean free path length 31 see also probes electron rest mass 825t

electron scattering, elastic/inelastic scattering processes in TEM 298, 589–92, 619–23 electron scattering amplitude/factor 416, 483–4, 503, 507, 540 electron sources 283f energy distribution of electrons, full width at half maximum (FWHM) 285 Schottky emission 558–9 thermionic and field-emission sources/cathodes 284t, 558–9 electron spectroscopy for chemical analysis (ESCA) 135, 145 surface analysis 119–20 electron tomography (CAT) 650–5f in biology 653–4 with EFTEM 651–5 electron transitions dipole selection rules 78–80, 85, 88f, 100 electron binding energies 79t lifetime 84 mixed state 84 particle-induced X-ray emission (PIXE) 78 transition probability between states 84, 85f two-transition processes 78 electron volt (eV) defined 24 electronegativity 155, 826f, 131 electronic structure 57 elements 74t, 75f hybridized orbitals 76, 77f overview 57–8 spectroscopic techniques 186f valence s and p energy levels 76f electrons as probes 294–303 backscattered electron current 70–10

backscattered angular dependence 715–16f beam broadening 294 beam heating 301 differences in scattering, X-rays and fast electrons 505 elastic interactions 589 elastic/inelastic collisions 294–5 free electron gas 31 interaction with atom resulting in core hole 35f intrinsic spin magnetic moment 76 irradiation damage 298 K-shell 35 kinetic energy 26, 31f mass 25, 526t mean free path of 31f, 295 neutrons and X-rays, scattering factors 536t outgoing and incident currents 709 photoemission 97 secondary spatial distribution 297 sputtering 300 in TEM, wavelength 505 thermionic and field-emission sources 282–4t total electron yield 709 electropolishing system alloys 666 double-jet 667f electrostatic charging 302 electrostatic electron analyzers 123f electrostatic systems 290 elements binding energy, atomic number effect (Z) 131f Cromer-Mann atomic scattering parameters 421t electron binding energies 79t, 87f energies of X-ray emission lines 89t

OUP CORRECTED – FINAL

832 Index elements (cont.) ionization energy (eV)/electronic structure 74t KLL Auger lines 94t mass absorption coefficients and densities 103t, 104f periodic table 335f, 826f plasmon loss data at 100 keV 627t positive ion yield as a function of atomic number 314f stopping power for protons and He ions 317t ellipsometry 392–400 amplitude ratio 399 compensator (retarder) and photoelastic modulator 397 measurement principle 398–9f optical components used 397 optical microscopy 392–9 spectra for SiO2 399f summary 814t elliptically polarized wave 28f, 353–6f emission profile, mercury arc lamp 279 energy (eV) conversion to wave-vector (k) scale 205 energy intensity, unit 28 energy level diagram, transitions allowed by dipole selection rules 78–80, 85, 88f, 100 energy—displacement curve, molecular vibration 158 energy-dispersive X-ray spectrometry (EDXS) 15, 22, 78, 110–12f, 140 compared with wavelength dispersive spectrometry (WDS) 109f, 112 detector 18f, 110–11f, 140, 617–18f, 642f –3, 695f windowless 660 microanalysis 146, 642f

quantitative chemical mapping 56 spectrum 642f k-factors 118, 642f, 644f, 646, 727 summary 806t energy-filtered transmission electron microscope (EFTEM) 637–41, 639f biological imaging 654–5 image resolution of 1 nm 660 Mn(PdMn) multilayer specimen 641f tomography 653–4f vs STEM-EELS 640f energy-loss spectrum basic edge shapes, core electrons 629 ideal 633f near edge structure (ELNES) 621, 629, 630, 805t engineering materials failure 714–15f engines GE 90–115B jet engine 9f high-efficiency, MSE 3f ultrahigh temperature materials 8–10 enstatite 16 environmental scanning electron microscopy (ESEM) 732–4f, 738 epitaxial behavior, insulator– semiconductor 195, 195fn epitaxy 7fn, 222 escape depth, photons 22, photoelectrons 31f Escherichia coli chaperonins, GroEL and GroES 787 dispersed on mica 790f etchants 389f, 390t, 668 experiment, polymethyl methacrylate (PMMA) 707f eucentric height 554 Eulerian cradle, four-circle goniometer 442f –3, 446

eutectic Cu–Al alloy 718f eutectic liquid 18, 18fn Everhart–Thornley electron detector 698f –699 Ewald, PP 259fn Fifty Years of X-Ray Diffraction 257fn Ewald sphere construction 263f, 435–6f, 438f, 455 Bragg diffraction 259–60f, 489 comparison of X-ray and electron diffraction 264f for beam broadening 436f for powder or polycrystalline specimens 441f for surface diffraction 488f for LEED 491f for RHEED 495f, 497f for transmission electron diffraction 507f four-circle goniometer 442f –3, 446 Laue criterion 440 low energy electron diffraction (LEED) 491f, 548 reciprocal lattice rods 495f two different conditions of excitation error 522f in two dimensions 487–8 excitation error 519, 522, 524 exciton, defined 198fn exclusion principle (Pauli) 72 exit surface specimen 300, 505 exsolution lamellae 16 extended energy-loss fine structure (EXELFS) 621, 630, 805t extended X-ray absorption fine structure (EXAFS) 199–206, 210, 621, 630, 812t external magnetic fields, ferromagnetic specimens 714 extinction coefficient K, complex refractive index 392–3 extinction distance 513

for a specific diffracted beam (or reflection) 503 extreme UV (EUV) radiations 22f, 23 eye, human 375f –6

F face centered cubic (FCC) structure 20–1f, 227f, 233–5f, 428t, 523f, 548f crystals, perfect dislocations 658t Ni-based superalloys 8 reciprocal lattice 253f single crystal spot transmission electron diffraction patterns 517f structure factor 425 failure analysis 7 engineering materials 714–15f metallurgy of RMS Titanic 13–15f semiconductor devices 660–4, 664f Fe nanoelements, magnetic array 784f feldspars 16, 17f Fermi level 78, 189, 757 Fermi–Dirac distribution 151–2, 193, 195f ferromagnetic 3d transition metals, L-edge XAS spectra 201 ferromagnetic specimens, external magnetic fields 714, 737 Fibonacci sequence 266fn field emission gun (FEG) 558–9, 694 field ion microscopy (FIM) 46f –7 field-emission sources/cathodes 284t, 558–9 Fifty Years of X-Ray Diffraction, Ewald et al. (1962) 257fn fine-grain aggregates 16 fingerprinting, with frequency modulated atomic force microscopy 782f

OUP CORRECTED – FINAL

Index 833 first-order Laue zone (FOLZ) ring 650 fission reactors and spallation sources 287 fluorescence effect (F) 117 fluorescence optical microscope 655 FluoroGold labels 655fn fluorophores 280fn focused ion beam (FIB) H-bar and lift-out methods 670–1f milling 666, 669–70 SEM 734–5f top-down nanofabrication 738 focused probe illumination, TEM 566–7 force constant A, bond strength 158 force—distance curves, AFM cantilevers 790–1 form, size and dimensionality 2 Foucault, Jean-BernardLeon 609fn Foucault mode, domain imaging 610, 611f four-circle goniometer, Eulerian cradle 442f –3, 446 Fourier cosine transform of intensity 176 Fourier transform IR (FTIR) spectroscopy 30, 173–5f, 209f application in aircraft manufacturing 178f compared with Raman and NIR methods 185f Michelson interferometer 173, 218 practice and application 177–9 protein structure 206–7 summary 807t Fourier transform Raman spectroscopy 182, 205 Fowler–Nordheim formula 285 Fraunhofer, Josef 362fn Fraunhofer diffraction 346, 363–9, 432

from double and multiple slits 366 from single slit 363–5 relationship to X-ray and electron diffraction 368 free electron gas 31 frequency modulated atomic force microscopy (FM-AFM) 774, 775f, 782f Fresnel, Augustin-Jean 362fn, 609fn Fresnel diffraction 362f, 370–3 circular aperture 373f sagittal formula 371 Fresnel equations 398 Fresnel fringes 606fn Fresnel images 610, 611f Fresnel zone plate lenses 42, 373–4f friction force microscopy (FFM) 772–3 Friedel law 431 front focal plane (FFP) 584 full-field soft X-ray microscope 43 functional groups, Raman frequencies 183t

G GaAs band structure 156f crystal structure 237f z-contrast image 599f scanning thermal microscopy 785f gamma correction, image processing 52 GaN semiconductor, cross-section specimen 736 GaN ultra-thin electron-transparent specimen 735f garnetiferous metamorphic rocks 17f gas deposition,FIB 736 gaseous pollutants, Pt-based three-way catalyst (TWC) 208 Gatan image filter (GIF) 67 gate dielectric structures 662f breakdown 664f –5

nm-scale gate oxide 661 poly-Si gate structure 661f replacing SiO2 663 TaC gate and poly-Si cap 663 Gaussian image plane (GIP) 563 GdBa2 Cu3 O7 electron energy-loss spectroscopy (EELS) 624f Ge detector 112f through focus series images of amorphous Ge 583f generalized oscillator strength (GOS) 628 Glan–Taylor prism 397–8f glassy or amorphous materials 8, 39f, 223f glide planes 431 gold Au islands on a C film 586 back-scattering yield for 1.4 MeV 4He+ ions 321 FluoroGold 655fn golden ratio 266fn grain boundary phases 17–19 bright vs dark field imaging 576f bright-field TEM image 19 grains 222 size, X-ray peak broadening 433–7 graphite bonding in 77f displacement energy 299t electron-yield in C KVV Auger electron spectra 127 X-ray absorption spectrum 218 gratings see diffraction gratings ground state 73, 700

H half-period zone 371–3 Hamaker constant 766, 772 harmonic motion 349fn harmonic oscillator 158 classical vibrational frequency 159

Hartree–Slater (HS) wave-functions 636 4 He+ ions ions penetrating a solid 311 back-scattering spectra 40, 319f, 321f back-scattering yield 322f scattering 311f sensitivity to neutralization 326 stopping power S 315, 316f, 317t as a function of incident ion energy 316f heat assisted electromigration 758 Heisenberg, Werner 70fn helical structures, X-ray diffraction (XRD) photograph 11f –12 helium, stopping power for select elements and energies 317t HeNe laser 281 Hertz model in SFM 770 hexagonal close packed (HCP) structure 233–4f, 254, 428–9, 517f single crystal spot transmission electron diffraction patterns 517f, 518f hexagonal crystal system, Miller–Bravais indices 229–30 HfO2 film, on Si 662 HfZrOx dielectric structures 662 high angle annular dark-field (HAADF) detector 555, 653 HAADF–STEM image 599f Z-contrast image 601f high angle incoherent scattering, Z-contrast imaging 598–9 high energy electron diffraction 26, 482–5 see also reflection high energy electron diffraction (RHEED)

OUP CORRECTED – FINAL

834 Index high voltage electron microscope (HVEM) 33, 562 high-pass filters 53 high-resolution electron microscopy (HREM) 19, 42, 562, 601–5 nm-scale gate oxide 661–2f phase contrast image 559–600f, 601–5, structure images 606, 607 summary 819t higher-order Laue zone (HOLZ) lines 523f, 526, 548 histogram clustered 51 of intensities 56f histogram equalization 51 HOMO–LUMO energy gap 164f Hooke, Robert 345 Hooke’s law 770 Howie–Whelan equations 513 human eye 375f –6 peak sensitivity 347 Hund rules 148 Huygens, C 357fn Huygens’s principle 356–7f, 371 hydration of cement 538–40 time-resolved neutronscattering 538–40 hydrocarbon contamination 303, 712fn hydrogen atom, Bohr radius 23, 70, 591, 620 bonding 149f amide groups in protein 207 protons, stopping power S 315, 316f, 317t quasi-elastic scattering 539f time-resolved neutron-scattering 539 zero point energy 159 hydrogen/deuterium 159 protein structure and 159 hydrogenated diblock copolymer (PEO-PB) 775 hydroxystilbamidine 655fn

hysteresis and creep, function of applied voltage 777

I icosahedrite, point group symmetry 267f image acquisition software 51 image contrast and noise 706–7 secondary and back-scattered electrons 709–10 TEM 555 image segmentation 55–6f imaging 32, 818t–3t digital imaging 48–56 high-pass/low-pass filters 53, 55f microscopy 41–8 one-dimensional structure images 603f optical imaging 375–401 point, neighborhood, kernel operations 49–55f recording and detection of electrons 588–9 see also objective lens (OL) imperfect crystals, two-beam diffraction condition 509–14 in situ methods 56–9 in situ TEM 665 X-ray diffraction 468 Neutron scattering 53840 incident radiation, scattering 40f incoherent scattering 504–5 Compton 413–14 mass—thickness contrast 592 induction-coupled plasma mass spectrometry (ICP-MS) 278, 331–6f optical detection system (ICP-OES) 333–4t in pharmaceutical industry 334 summary 808t inelastic scattering 166, 190, 294–6f, 298f, 504–5, 617–18f, 619–23

and beam broadening 707–9 cross-section 622f EELS 557, 623–32 inner-shell (KLM, N and O) ionization 621 Kikuchi lines 519–520f spectroscopy 619–23 infra-red spectroscopy 5t,147f, 161, 172–5 complementary transmittance 170f components 174t Fourier transform (FTIR) 174–9 Michelson interferometer 174–175f Near IR, FTIR and Raman comparison 185f instrumentation 173–6 Raman activity, selection criteria 169–70 see also Fourier Transform Infrared Spectroscopy (FTIR) ink-jet printer 667 instrument artifacts, SPM 780 insulator—semiconductor, epitaxial behavior 195 integrated circuits (ICs) 660 intensity, Fourier cosine transform 176 interaction cross-section 589 interaction forces and contrast formation in SFM 768f interatomic bonding in solids 147, 152–7 covalent 154–5f ionic 153–4 interface energies 498 interference 2-slit diffraction grating 38f interference contrast microscopy 383 interferometry beam splitter 175 interferogram 173 Michelson 174, 175f, 348 polystyrene 177 range of frequencies 178

technique, SFM 763 intergrowths 16 intermediate voltage electron microscopy (IVEM) 33fn intermittent contact AFM (IC-AFM) 759, 774, 774f, 789 intermittent contact/dynamic/tapping modes 759 International Center for Diffraction Data 441 International Tables for Crystallography (ITC) 239–44, 240fn, 268 space group Pnma 268 interplanar spacing dhkl 230t, 427, 527 inverse photoemission spectroscopy (IPES) 151, 155–6f, 187, 191f –2 summary 808t ion beam milling 664, 666, 669 ion beam sputtering 312–14, 313 ion channeling 324 determination of foreign atoms in lattice 325f ion implantation 312 ion milling, dimple grinding 667 ionic bonding 153–5 schema 153f ionization, secondary ion yield 314 ionization cross-section Q 621, 635 ionization energy (eV) 74t ions channeling 324–5 cross-sections (barns) 305 dimension, sizes and cross-sections 304 excitation (inelastic scattering) 308 generation of protons (H+) and alpha (4He+) particles 291 guiding and focusing 288–90

OUP CORRECTED – FINAL

Index 835 interaction of high-energy ions with materials 32, 306f ion implantation 307 ion-based characterization methods 315–16 Kinematic collisons, physics 307–10 positive ion yield as a function of atomic number for elements and compounds 314f primary ions 305 sources 291–2 stopping power, protons and 4He 317t very high-energy (MeV) primary ions 305 see also secondary ion mass spectrometry (SIMS), Low-energy ion Scattering Spectroscopy (LEISS) and Rutherford Backscattering Spectroscopy ion–solid interactions 305–6f iron alloys, X-ray transmission microscopy (XTM) 43f iron meteorite, Toluca 41–2f iron oxide nanoparticles 528f, 602f dispersed on a carbon grid 602f superparamagnetic (SPIONs) 594–5 irradiation damage, electron 298

J Joint Photograph Expert Group (JPEG) 50–1f jump off contact, SFM 769 jump to contact, SFM 768 jump-ratio imaging, EFTEM 639, 654

K K-beta radiation 100 filters for suppression 104t

k-factors 644f –7 energy-dispersive X-ray spectrometry (EDXS) 644f, 646 microanalysis using 118 K-line, all principal elements up to Cu 644f K-shell 86–8, 100–1, 188 kaolinite 16 kernel/neighborhood operation 49–55f kernels 49–50, 53, 55f partial x- and y-derivatives 55f Kikuchi lines 518–23, 541 construction 523f maps and patterns 518–20 tilting to exact Bragg orientation 522f kinematic collisions 305–7, 308f kinematic factor, ion collisons 309, 310f, 319 kinematic scattering, electrons column approximation 541 geometry 308f kinematic theory 422, 485 electron diffraction 506–9 kinetic energy, electrons 26, 31f Knoll, Ruska 584fn Kossel cones 520f –1 Kossel patterns 524f beam convergence semi-angle 524 krypton, electronic structure 72

L L- and M-radiation 100 L-edge XAS spectra 201 L-shell 188 lamps 278–80 tungsten-halogen 279f mercury-arc 279f lanthanum hexaboride (LaB6) thermionic emissions 186, 283f work function 286 lanthanum oxynitride (La5 Si3 O12 N) 20 laser ablation ICP-MS 332

laser scanning tunneling microscope (LSTM) 748f lasers 280–1f common gas and solid-state lasers 282t emission wavelengths in nm 282t HeNe laser 281 laser-induced mass analysis (LIMA) 294 light sources 181–2 pulsed-laser irradiation 666 Raman spectroscopy 181 ruby solid-state laser 281f tunable laser 182 UV 182 lasing 280 lateral force microscopy (LFM) 772–3f lattice fringes 601–3 phase contrast imaging 603 two-dimensional image 604 lattice nets 486f two dimensions 485 reciprocal 487–8 lattices 223–4, 250–2 anisotropy and channeling effects 708f Bravais lattices 224–6f, 225t, 236 crystallite or grain, strain 460 determination of the position of foreign atoms 325f parameters, measurement of 460 points, lines, directions, and planes 227–9 reciprocal lattice 155, 250, 250–6f, 251–6f, 268 space lattice 223–4f strain measurements, crystallite or grain size 460 translations and rotational axes 266f vibrations (phonons) 160 Laue, Max von 256fn Laue criterion of diffraction 256, 423, 435, 507 Ewald sphere 440

modified 507–8 Laue groups 431 Laue zones 263, 649 higher-order Laue zone (HOLZ) 523f, 548 zero order Laue zone (ZOLZ) 263, 525f lead–barium–titanate (PBT) or lead–zirconium–titanate (PZT), piezoelectric ceramics 777 length scales atomic level 2 macroscopic/mesoscopic/atomic levels 2 lens aberrations, correction 378, 587–8 distortions and astigmatism 378–9, 605–6 human eye 375f –6 TEM, intermediate lens astigmatism 514fn see also electromagnetic lenses lens excitations 515f Leonard–Jones potentials 765 Lewis, GM 24fn light advanced light source (ALS) 106fn, 107f sources, lasers 181–2 velocity 825t light microscopy see optical microscopy limiting sphere construction 260f Linac 287–8 linear combination of atomic orbitals (LCAO) 149 linear particle accelerator (Linac) 287–8 linear voltage differential transformer (LVDT) 790, 792, 793 lipid membranes, atomic force microscopy (AFM) 789f, 790 local density of states, mapping 747 local electrode atom probe (LEAP) 46

OUP CORRECTED – FINAL

836 Index long-period multilayers 531–2 long-range order (LRO) defined 223 long-range oscillations (EXAFS) 199f, 202 long-range tip–specimen attractive force 770f look-up table (LUT) 50–2f gamma correction 52f Lorentz, Hendrik Antoon 609fn Lorentz force 98, 560, 610, 630 beam deflection coils 697f centripetal force 630 helical trajectory of electron 561f secondary electrons 723 Lorentz microscopy (TEM) 43, 43f,609–10 summary 820t Lorentz polarization factor 455–7, 459t low energy electron diffraction (LEED) 26, 482–5, 490–5, 539, 548, 551 apparatus 490–1f Ewald sphere construction 491f summary 815t low energy ion-scattering spectroscopy (LEISS) 278, 308, 310, 315, 325–8 summary 809t low-pass filters 53, 55f luminescence 700 intrinsic/extrinsic 732 lysostaphin 791

M M-radiation 100 Madelung, Erwin 154fn Madelung constant 154 maghemite/magnetite Fe3 O4 528 magnetic alloy, Nd2 Fe14 B, SAD patterns 518f magnetic array, Fe nanoelements 784f magnetic contrast types I/II, SEM 722 magnetic fields (H) 26

magnetic flux quantum 825t magnetic force microscopy (MFM) 766 improved resolution 784f involving two scans 783f –4 spatial resolution 784 tapping mode 784 magnetic interactions, neutron scattering 537 magnetic permeability of free space 825t magnetic prism 630 magnetic quantum number 72 magnetic resonance imaging (MRI) 783 magnetic skyrmion lattice 43 magnetization ripple 609fn, 612 magnetooptic Kerr effect 282 magnets bending/undulators/wigglers 105–6f electron holography 613–17f monoatomic ferromagnets/ antiferromagnets 538 MSE 3f specimen-collector geometry 722f synchrotron radiation 106f in synchrotron rings 105 mammalian cells, atomic force microscopy (AFM) 790 Mars geology, X-ray diffraction measurements 469–71 martensite 534 mass absorption coefficients and densities, select elements and X-ray radiations 102t, 104f mass spectrometry quadrupole mass spectrometer 330f time of flight (TOF) mass spectrometer 330–2f see also induction-coupled plasma mass spectrometry (ICP-MS) mass—thickness contrast 592 incoherent elastic scattering 594

mass—thickness scattering contrast 591 materials characterization 1–2, 57–8 core activities tetrahedron 4 examples 7–8 societal impact 3f summary 59 materials science and engineering (MSE) societal impact 3f matrix effects 312 matrix factors (ZAF) 116 atomic number effect (Z), absorption effect (A), fluorescence effect (F) 117–18 Maxwell, James Clerk 27fn Maxwell equations 27 Maxwell–Boltzmann distribution 286 mean free path length 30–31f, 124, 305, 485 plasmon excitation 627 scattering cross-section 590–1 total, in vitrified ice 654 mechanical loop, defined 776 medical specimens, X-ray fluorescence spectroscopy (XRF) 137–8 Meitner, Lise 91 Meniscus, formation 766 Mercury, emission characteristics 279f mercury arc lamp, emission profile 279–80 mesoscopic/microscopic level, length scales 2 metal catalysts 208–9 metal oxide semiconductor field effect transistor (MOSFET) 660 metallurgy, RMS Titanic 13–15f metals 3d transition `white lines’ 200 bonds/bonding 156–7 calculated spin-split density of states 157 displacement energy 299t

electrochemical polishing 667 failure analysis of the RMS Titanic 13–15f hexagonal close packed (HCP) structure 233–4f, 254, 428–9, 517f impurities detected by ICP-OES 334 mean free path of electrons 31f metallography 389–92 metallography etchants 391–2 metallography/optical microscopy 41 photoelectron spectra 152 metal–halide optical source 279f meteorite, Toluca 41–2f metrology 7, 660, 781 MgAl2 O4 crystal (spinel) 220f, 238, 525f MgO films 447 mica 16 Michelson interferometer 174, 175f, 348, 763 micro-focused X-ray fluorescence system 138f microanalysis, minimum mass fraction (MMF), EDXS microanalysis 645 microdiffraction 648–50 elastic scattering 648 fine probe incident beam 648 micromachining, FIB 736 microscope, optical see optical microscopy microstructure dual-phase steel alloys 20–1f rocks and minerals 15–17 structural/chemical/morphological aspects 21 microtome, ultramicrotomy 668f, 669–70 microwaves 23f Miller, WH 228fn Miller indices 252 crystalline materials 228–9

OUP CORRECTED – FINAL

Index 837 Miller–Bravais indices, hexagonal system 229–30 minerals, microstructure 15–17 minimum detectable mass (MDM), EELS microanalysis 636 minimum mass fraction (MMF) 645 Mn K-edge EXAFS spectrum, photosynthesis 206f Mn—Pd ordered tetragonal structure 531f Mn2+ and Mn4+ L3,2 edge 630f Mo characteristic X-ray emissions 87–8 Mo–Mo distances, R 205 Mo–Si bilayers 532f Mo X-ray tube, X-ray intensity as function of wavelength 99f Mo—Si thin films and multilayers 658–0f comprehensive TEM analysis 659f Mo-K edge, absorption spectrum 203–4 moire patterns 608 molecular energy levels 148–61, 208 band gaps in semi-conductors 152f density of states (DOS) 150–1f electronic/vibrational/rotational contributions 157, 158f, 160f, 161f, 208 molecular orbitals highest occupied (HOMO) 150, 150f, 163 HOMO–LUMO energy gap 164f larger molecules 149–50 linear combination of atomic orbitals (LCAO) 149 lowest unoccupied (LUMO) 150, 150f, 163 molecular spectra 158–72 Rayleigh and Raman scattering 166–72

vibration and rotational modes 158–162 monochromators 198, 450–1f Monte Carlo simulations electrons propagating in Al 708 low/high average atomic number 592–4f MOSFETs, metal—oxide semiconductor field effect transistors 660, 785 Mosley diagram 145 motor exhaust, Pt-based three-way catalyst 208 Mott detector 738 Muller, ER 47fn multilayers diffractometry, reflectivity and pole figures 445–9f energy dispersive X-ray microanalysis 660 see also thin films and multilayers multiplicity (planes) 455, 459t muscovite 16, 17f M×N surface reconstruction 487

N NaCl annotated PDF card for NaCl 462f ionic bonding 153–4f powder X-ray diffractometry data 441f structure 236–7 NaH, rotational transition 162 nanocrystals, semiconductors 165 nanoelements, magnetic array 784f nanofabrication, top-down focused ion beam (FIB) 738 nanolithography, dip-pen (DPN) 767, 793–4 nanomanipulator 671 nanoparticles 8 arrays 527 crystallographic relationships 8 iron oxide 528f, 602f

specimen preparation for TEM 666 superparamagnetic iron oxide nanoparticles (SPIONs) 594–5 nanoscale structures, gas deposition 736 nanoscience 2 naphthalene, UV-Vis 165f 4 He+ ions, sensitivity to neutralization 326 near-edge fine structure X-ray absorption spectroscopy (NEXAFS) 199–206, 210, 621 nearest neighbor distance 203 neighborhood/kernel operations 49–55f neutron scattering 31, 303, 337 fission reactors and spallation sources 287 hydration of cement 538–40 in situ kinetic studies 538–40 interactions with matter 535–41 magnetic scattering amplitude 537 nuclear interactions 535–6 nucleus–neutron “complex” 536 penetration depth 31, 337 polarization 538 scattering factor 535, 536 scattering geometry from a magnetic sample 538f summary 815t thermal 539–40 Newton’s law of motion 71 Ni, thin films grown on silicon 321–2f Ni alloys, comparison of secondary and back-scattered SE micrographs 720f Ni-based superalloys 8–10 addition of Re 10f dual-phase microstructure 8 single crystal superalloys 10f ultrahigh temperature materials for jet engines 8

Ni-silicide 664f NIR spectroscopy, vs Raman and FTIR methods 185f noble gases 78 4 Ar+ ions, sensitivity to neutralization 326 electronic structure 72 Kr electronic structure 72 manipulation of adsorbed Xe atoms on clean surfaces 757 sensitivity to neutralization 326 Xe emission characteristics 279f see also helium non-cubic materials, crossed polarizers 42 nuclear charge +Ze 69 nuclear magnetic resonance (NMR) 22, 69 nuclear reaction analysis (NRA) 305 nucleus–neutron “complex” 536

O Oak Ridge National Laboratory (USA) spallation sources 287–8f objective aperture (OA) 568, 593 objective lens (OL) additional phase shifts 578 back focal plane (BFP) 585f increasing depth of field 697 optical microscope 381–388 phase contrast imaging 577 SEM 598 TEM 566–70 occupied states 751 1-octadecanethiol (ODT) 794 octahedral sites 235 Omicron-HAXPES system 122f One Ångstrom Microscope (OAM) 45 one-dimensional structure images 603f

OUP CORRECTED – FINAL

838 Index one-electron transition picture, p - d transition 201f operations see imaging operations optical beam deflection 763 optical fibers 3f optical imaging 375–401 optical microscopy 376–91 Abbe criterion 376–7 confocal scanning 386 direct, oblique, and dark field imaging 383 ellipsometry 392–9 geometric optics and aberrations 378–83 improving resolution 44 interference contrast microscopy 383–4 light sources 278 metallography 41, 389–92 polarized light 384 Rayleigh criteria 376–7 summary 820t vertical illumination in reflection geometry 381–2 optical spectroscopy 37fn absorption 152f, 186f vibrational/rotational modes 158–9 optics and optical methods 345–408 orbital quantum number 72, 75f orbiting nuclear charge +Ze 76, 77f orientation relationship 549 electron diffraction patterns 535, 549 crystallographic phases in Sm-Co alloys 529–530f epitaxial growth 447 twin boundary 249, 532–5 orthoclase 16 oscillating or accelerating electric charge 58 oscillation amplitude, AFM 774 ‘oscillator’ electron 504 oxygen-evolving complex (OEC) 205

P p - d transition, electron picture 201f paintings, X-ray fluorescence spectroscopy (XRF) 12–13, 68f parallelepipeds crystal 223 paraxial ray 379fn ‘particle in a box’ model 758 particle-induced X-ray emission spectrometry (PIXE) 78, 278, 336–7 air pollution 336f electron transitions 78 K-alpha spectrum from Ti 336f summary 809t Pasteur’s quadrant 2 path difference between diffracted and forward scattered beams 504 Pauli, Wolfgang 72fn Pauli exclusion principle 72 Pauling electronegativity 155, 825f Pb frequency, modulated atomic force microscopy (FM-AFM) 782f peletic schist 17f perfect dislocations, in FCC crystals 658t performance characterization, MSE 3f periodic table of elements 826f detection limits in ICP-MS 335f perovskite, structure 238 pharmaceuticals impurity detection by ICP-OES 334 phase angle 347–8 phase contrast imaging 577 Al-O-N system 603f contrast transfer function 575, 579–80f differential (DPC) 511–12f, 610 high-resolution electron microscopy (HREM) 601–4, 659f –60

lattice fringes 602 weak phase object (WPO) approximation 577 phase difference, elastic scattering 506 phase factor, arising from spatial distribution 483 phase identification, and structure refinement 462 phase object approximation (POA) 577 phasor representation and addition of waves 350–1 phonons 160 defined 452 phosphorescent radiation 280 photo multiplier tube (PMT) 588 photo yield method in XAS 199 photo-detector 108 photodiode detector 760f, 763 photoelasticity, birefringence behaviour 398 photoelectric effect 24 photoelectron emission 25, 35f, 97, 131 cross-section calculated for different sub-shells 136f mean escape depth 199 photoelectron energy spectrum 217f photoelectron waves, EXAFS 203f photoemission microscope with scanning aperture (PEMSA) 748f photoemission spectroscopy (PES) 128–9, 152, 155–6f, 188–197 angle-integrated spectra 196 angle-resolved 155, 188, 197 binding energies 195 chemical identification and analysis 196 compared with XAS 199 inverse (IPES) 151f, 155–6f, 187, 191f –2 photoelectron spectra 152 photon sources 129

summary 208–10 photoionization event 130 photons 35, 37 absorption depth 199 cathodoluminescence 700f, 731–2 damage 33, 293–4 defined 29f electric field component E (z) 29f electromagnetic spectrum 25 escape depth 22 incidence 25, 113 Lamps and lasers 278–82 penetration depth 22, 30, 293–4 probe, constant energy 188 soft X-ray 293–4 sources 129, 278 two-photon process 168 UV-Vis spectroscopy 162–6 as wave packets 29 wave particle duality 24– X-ray fluorescence spectroscopy (XRF) 113–8 X-ray photons 91, 98–112, 413–4 photosynthesis Mn K-edge EXAFS spectrum 206f oxygen-evolving complex (OEC) 205 Picasso painting, X-ray fluorescence spectroscopy (XRF) 12–13f, 69f piezoelectric actuators 747 piezoelectric ceramics, lead– zirconium–titanate, lead–barium– titanate 776 piezoelectric materials 747fn piezoelectric scanners 778f pixel 48 Planck, Max 24fn Planck’s constant 24, 25, 29, 70, 825t reduced 71, 825t plane polarized electromagnetic wave 27f

OUP CORRECTED – FINAL

Index 839 plane waves, linearly or plane polarized 27, 356f, 384–6 plasma frequency 31 and loss function 31 plasma ion source, and duo plasmatron ion source 291 plasmon excitation 618f, 626–7 cross-section 620 path length 626 plasmon loss data at 100 keV for some elements 627t EELS 625f, 627 plasmon resonances 297 plasmon scattering 626 cross-section per atom/m2 621 plasmon scattering angle 620 plasmons conduction electron bands 620–1 mean free path 626 platinum, Pt-based three-way catalyst (TWC) 208 pneumoconiosis 137 point groups 225, 241 point incident beam, broadening trajectory envelopes 708f point operations, digital imaging 51 Poisson ratio 657, 771 polar diagram 520f polar intensity plot, electron beam 715 polarizability, proportionality constant 167–8 polarization factor, Lorenz 455 polarized light circularly vs elliptically 354–6, 397 linearly 353, 354–6, 397–8f, 413 optical microscopy 384–6 p- and s-polarized light waves 394–6 polarizer (analyzer) 397 pole figures 445–9 pollutants, Pt-based three-way catalyst (TWC) 208

poly-Si gate structure 661f polyatomic molecules, linear combination of atomic orbitals (LCAO) 149–50f polycrystalline film, diffraction 446–7 polycrystalline specimen, backscateered image 717f polymer-matrix composites, FTIR 179 polymers, curing 177–9 polymers and biological tissues, specimen preparation 667–8f polymethyl methacrylate (PMMA) etching experiment 707f polystyrene, FTIR 177f polytype structures 603fn, 603f population inversion 280 position-sensitive detector 632f powders and nanoparticles Debye–Scherrer camera/ method 443–5f, 458t diffraction data 458–9t diffraction files (PDFs) 441, 461–2, 529 phase identification and structure refinement 461, 462 and single crystals, diffractometry 438–41 specimen preparation 667 principal quantum number 72, 73f prism electron biprism 556, 556–7f, 613, 614f Glan–Taylor prism 397–8f magnetic prism 630 probability distribution, angular component of 75f probe diameter and probe current 695 probe radiation 292 probes 278–338 artifacts in SPM 777, 779f based on electromagnetic spectrum 22–3

cantilevers 759–65, 761fn, 785f characterization methods 5t classification as diffraction/spectroscopy/ imaging 4 criteria for technique selection 29–31 diffraction limited diameter 702 electrons 282–7 electron probe effective diameter 701 formation 697f ideal probe diameter 702 incident probe size and spatial resolution 701 interactions with matter 29–34, 292–315 short-range 747 ions 288–92 mean free path length 30, 31f, 305 neutrons 287–8f penetration depth 22, 30, 293–4 photons 278–82 SEM and TEM 30 and signals 22, 59 spectroscopy, diffraction and imaging 34–6 summary 59, 337–8 tip, short-range interaction 747 X-rays, 97–107 process development, semiconductor 660–6 processing–structure– property–performance tetrahedron 4f projector lenses 554 propane in hexane solvent, UV-Vis 165f proportionality constant K 250fn, 252fn polarizability 167 protein structure amide groups, hydrogen bonds 207 characterization 159 deuteration 207f FTIR 207–8

protein—DNA complexes, atomic force microscopy (AFM) 787–8f protons 303–4 cross-sections as a function of atomic number 305f interactions with matter 303–4 penetration depth 337 PIXE 336f rest mass 825t sources 291 stopping power, for select elements and energies 317t Pt-based three-way catalyst, automobile exhaust 208 pyroxenes 16 microanalysis using stoichiometric oxide standards 742t

Q Q factors 774 resonance behavior of oscillator 761fn quadrupole lens array, EELS 632f, 639f quadrupole mass spectrometer 330f quantitative chemical mapping, EDXS 56 quantitative X-ray microanalysis 116–19 with energy dispersive X-ray spectrometry 641–8 quantization, image processing 49 quantum mechanics 758 model of the atom 69–78 scanning tunneling microscope (STM) 746, 750–2 transition probability 84 tunneling effect 284 vibration modes 159–160f quantum numbers 758 spin 72 magnetic 72 orbital 72 Pauli exclusion principle 72 principal 72, 73f

OUP CORRECTED – FINAL

840 Index quantum numbers (cont.) resonance Raman total angular j and mj 77, 85 spectrum 180f quartz, SiO2 16, 17f, 198 selection criteria quasi-elastic scattering, 169–70 hydrogen 539f summary 810t quasicrystals 266, 481f, 482 surface enhanced quasiparticle, defines smallest (SERS) 183–4 perturbation 43fn rare gases see helium; noble quenchers, inhibit or eliminate gases cathodoluminescence 732 Rayleigh, Lord 376fn Rayleigh criterion 32, 41–2f, R 561 defined 42f radiation for small angles 44 breaking 97, 101, 450 Rayleigh scattering 37, 168–9, characteristic X-ray 85–89t 210 damage, electron 298–300 Stokes/anti-Stokes dose 301 scattering 169 extreme (EUV) summary 210 radiations 22f –3 Rayleigh transitions 180f generating X-ray 98–101 reciprocal lattice 155, 250, incident radiation 40f 251–6f, 268 K-alpha, K-beta 85f, 100, Ewald sphere 104t construction 259f–65, plane polarized 27 507 probe 292 in two dimensions 487–9 scattering and nets 487–8 diffraction 37–41 points, broadening 433–6f synchrotron 105–6, 137–9, rods 495 204f surface diffraction 506 as a wave 24–26, 34 twinned crystal 532 white 439f unit cell, dimensions 505 see also electromagnetic —; vector 251, 507 infra-red —; reciprocity principle, synchrotron —; X-rays stated 585 radiative decay process 78, 84f reflection Raman, Sir CV 37fn, 147fn Fresnel equations of 394, Raman frequencies, common 396 functional groups 183t total internal 361 Raman scattering 37, 181 reflection high energy electron classical model 166–9, 210 diffraction filter and diffraction (RHEED) 26, 482–5, grating 181 494–501, 539 selection rules 170 Ewald sphere and reciprocal Stokes/anti-Stokes 166–9, 210 lattice rods 495f summary 210 possible RHEED scattering Raman spectroscopy 22, 147f, geometries 497f 161, 179–82 RHEED oscillations, in situ applications 182–4 monitoring of thin film compared with FTIR and growth 498–9 NIR methods 185f summary 816t first published Raman variation of intensity of spectra 37f specularly reflected instrumentation 181–2 beam 499f laser 181

see also high energy electron diffraction; Low energy electron diffraction; transmission electron diffraction refraction 359–61 complex refractive index N 392 Snell law 396 refractive index 361f relrods 509 resolution depth 32 Rayleigh criterion 41–2f spatial resolution 32 temporal resolution 32, 666 resolution scale 772 resonance behavior of oscillator 761fn resonance Raman spectrum 180 RF plasma ion source 291 Richardson, OW 282fn Richardson’s law 282 rocks and minerals, microstructure 15–17 Rohrer, Heinrich 47fn, 746fn Rontgen, Wilhelm C 23fn rotation axes, symmetry 240 rotation symmetry, tenfold 39f ruby solid-state laser 281f Ruska, Ernst 44fn, 553fn, 746fn Rutherford, Ernest 69fn Rutherford back-scattering spectrometry (RBS) 40, 278, 306, 315, 483 back-scattering energy-loss factor 320 comparison with LEISS 326F energy Width 318–19 kinematic scattering geometry 308–10 shape of spectrum 320–1 silicide analysis 322f summary 810t stopping power data, select elements and energies 317t

Rutherford cross-section 598 Rutherford scattering geometry of electrons, with atomic nuclei 483f model 589, 592f Rydberg constant 70

S Salah, Ibn 361fn sample and specimen 20fn saw-tooth shaped curve 621, 629 saw-tooth voltage, instrument artifacts 780 scan coils 568 scanning capacitance microscope (SCM) 748f scanning electron microscopy with polarization analysis (SEMPA) 700, 723–6f summary 821t scanning electron microscopy (SEM) 35, 44–6f, 114, 694–709 achievable resolution 704f back-scattered/secondary electron micrographs 719–20f beam-solid interactions and signals 699 beam broadening 707–9 cathodoluminescence 731–2 channeling and electron back-scattered diffraction patterns 720 collection of image information 553 column length 697–8 compositional maps, Cu–Zn specimen 46f depth of field, magnification, and achievable resolution 703–4f domain wall contrast 722f elastic and inelastic scattering and beam broadening 707 electron back-scattered diffraction patterns 720 Everhart-Thornley electron detector 698 environmental SEM 732–4

OUP CORRECTED – FINAL

Index 841 focused ion beam (FIB) column 694, 734–6 image contrast 706f, 709–20 imaging with different operating parameters 718 imaging magnetic domains 722–6 incident probe size and spatial resolution 701 instrument 694–8 key design principle 694 Mott detector 725f Noise and contrast 706–7 objective lens 695 with polarization analysis (SEMPA) 700, 723–6f, 821t probing sample composition and electronic structure 726–32 resolution 697, 702f achievable resolution 704f ultimate spatial resolution 701 scattering processes 296–7 SE and BSE detectors 717f secondary and back-scattered Ni alloy micrographs 720f signal radiation 694 spatial resolution 701–3 specimen preparation 736–7 stripe domains 722f summary 737–8, 821t type I and II contrast 722–3 two-condenser lens system 696f Topographical contrast 713–14 ultimate spatial resolution 701 variations 732–6 see also electron microscopy scanning force microscopy (SFM) applications 34, 781–93 artifacts 777–80 atomic-range forces 759 basic operation 767–77 cantilever function as force sensor 761f, 763–5

dynamic mode 767 instrumentation 776–7 interaction forces and contrast formation 768f interferometry technique 763 intermittent contact/dynamic/ tapping modes 759 lateral force microscopy 772–4 mechanical characteristics of cantilever 760–3 non-contact mode, dynamic 773–6 physics of 747, 759–67 principle 747 spatially resolved eigenstates of 2-D electron gas 759f static contact mode for topographic imaging 769–71 summary 822t tip and cantilever 760fn tip–specimen forces encountered 765–7 various deflection sensors 764f various designs 748f scanning ion conductance microscope (SICM) 748f scanning near-field acoustic microscope (SNAM) 748f scanning near-field optical microscope (SNOM) 748f scanning noise microscope (SNM) 748f scanning probe microscopy (SPM) 746–98 probe artifacts 777–8f tip artifacts 779f tip-surface interaction 746 scanning spreading resistance microscope (SSRM) 748f scanning thermal microscopy (SThM) 748f, 785–6

cantilever thermocouple probe 785f improvement 786 scanning transmission electron microscope(STEM) 584–6 optics, ray diagram 585f principle of reciprocity 583–4 vs conventional TEM (CTEM) 584f scanning tunneling microscope (STM) 46–7f, 500 attenuating barrier 749 barrier width 749 basic operation 752–59 constant current mode 753, 754–5f constant height mode 753, 754f elastic tunneling through a one-dimensional barrier 749–50 first published image of Si surface reconstruction 745 invention 746 with inverse photoemission (STMiP) 748f manipulation of adsorbed atoms on clean surfaces 757–9f physics of 746–7, 749–51 quantum mechanical tunneling model 750–1 quantum mechanics of tunneling 746–7f spectroscopy mode 755–7 summary 822t use 758 scanning tunneling potentiometry (SPotM) 748f scattering neutron 303, 484, 535–40 radiation as a wave 34 summary 210 scattering angle 310, 589, 620 scattering and diffraction 37–41, 58 incident radiation 40f

scattering and diffraction methods, summary 814t scattering factors, neutrons, electrons and X-rays 536, 536t scattering processes, TEM and SEM 297 Schawlow, AL 282fn Scherrer equation, peak broadening 435, 448, 460 Scherzer defocus and resolution 45, 580–3f,605 Schottky emission gun 45, 285, 553, 558–9, 661, 695–6 Schrödinger, Erwin 70fn Schrödinger equation 71–2, 160, 629 Screening, defined 591 screw axes 431 secondary and back-scattered electrons 698 secondary electron emission 710–12 angular distribution 712 effect of incident beam energy 711f effect of specimen tilt 712 secondary electron imaging depth of field 719f small bias field 713f surface irregularities 713f surface-tilt contrast 713–14 topographical contrast 713–14 secondary electrons, Lorentz force 723 secondary ion mass spectrometry (SIMS) 120, 278, 315, 328–30 summary 811t segmentation, Sobel edge detector 54f selected area diffraction (SAD) 482, 541, 573–4, 816t and BF/DF imaging in a TEM 573–4,

OUP CORRECTED – FINAL

842 Index selected area diffraction (SAD) (cont.) Debye–Scherrer type ring pattern 515f, 602 digital camera 514fn imaging 572f –3 ring and spot patterns 514–19 single crystal spot transmission patterns 518f semiconductors chemical polishing/etching 666 crystal structures 238, 244, 275 detectors 36 failure analysis 660–5, 664f MOSFETs 660–5 nanocrystals 165 Ni thin films grown on silicon 321 UV-Vis spectra of quantum dots 165f sensitizers or co-activators, promoting cathodoluminescence 732 shadow contrast, secondary electron and back-scattered images 714f shake-up peaks and multiplet splitting in XPS 134f shape transform 505 shear modulus 761 Shechtman, Dan 39fn, 221fn short wavelength limit (SWL) 99 short-range forces 765 short-range order (SRO)and diffuse scattering 467–8 short-range tip—apex—surface repulsion 770 Si bulk amorphous 663f diamond cubic crystal structure 518–19 frequency modulated atomic force microscopy (FM-AFM) 782

intensity distribution of secondary electrons 711f L-edge spectra of tetrahedrally coordinated Si 198f poly-Si gate structure 661f SiO2 quartz 16, 17f, 198 Si–Li and intrinsic Ge detectors 112f steps and various point defects 47f surface 7 x 7 reconstruction 745f Ta—Si film 128 topographical STM image 47f see also thin films and multilayers Si film, regrown on an amorphous SiO2 layer 721f Si nitride, sintering and grain boundary phases 17–19 Si—Mo, thin films and multilayers 658–60f Si3 N4 , sintering and grain boundary phases 17–19 Siegbahn, Kai 188fn SIGMAK and SIGMAL cross-sections 635 signal-to-noise ratio (SNR) 53 silica, gate dielectric structures 663 silicates mineralogy 16 quartz 16, 17f, 198 beta-SiAlON, structural determination 44 simple harmonic motion 346–8, 349fn phasor representation and addition of waves 350–1 simple harmonic wave 348 addition 400 addition of waves on orthogonal planes and polarization 353–4f complex representation 350–1 nodes/antinodes 352

standing wave 352 superposition of two waves of same frequency i351–3 but different amplitudes 356f simulated emissions 280–1f single crystal superalloy, microstructure analysis 9–10f sintering and grain boundary phases 17–19 skyrmion lattice 43 Sm–Co permanent magnet alloys 529–30f Sn frequency modulated atomic force microscopy (FM-AFM) 782 Snell law 396 Snellius, W 361fn Sobel edge detector 54f, 56 solids electronic structure 185–7f inverse photoemission 188 lattice vibrations (phonons) 160 photoemission methods 188–90 structure 185–7 Sommerfeld, Arnold J 69fn, 70 Sorby, Henry Clifton 41fn space groups in 2D 242f spallation sources 287 Oak Ridge National Laboratory (USA) 287–8f spatial resolution 32 specimen biological specimens 654 cross-section, GaN semiconductor 736 damage 33–4, 312 exit surface 505 handling and manipulation 570–1 side-entry/top-entry holder 570–1 TEM 666–7f layer-by-layer serial sectioning 736

Monte Carlo simulation in aluminum 708 and sample 20fn shape transform 505 standard 644 TEM holders 570–1 thickness amplitudes of different diffracted beams 605f measured by CBED 644 very thin 577 weak phase object (WPO) 577 specimen preparation 33, 666–8, 736–7 bulk samples 666 cross-section specimens for TEM 669–70 ion milling, dimple grinding 667 polymers and biological tissues 667 powders and nanoparticles 667 SEM 736–7 TEM 570–1, 666–71 top-down nanofabrication 738 specimen tilt back-scattered electrons (BSE) 716 secondary electron yield data 712f spectroscopy absorption, emissionand transition processes 34–6 and chemical methods 804t–13 electronic structure of solid 186f overview 57 with X-rays, core-level 107 see also angle-resolved photoemission — (ARPES); electron energy-loss — (EELS); Fourier transform IR (FTIR) —; inverse photoemission — (IPES); photoelectron — (PES); Raman —;

OUP CORRECTED – FINAL

Index 843 X–ray absorption —(XAS); X–ray fluorescence —(XRF); X–ray photoelectron —(XPS) specular reflection 499 sphalerite (zinc blende) structure 238, 244 spherical aberration 541, 559, 562, 581 coefficient 379, 562 electromagnetic lenses 541 spherically symmetry 483 spin polarized electrons, SPLEEM microscopy 26fn spin—orbit coupling, 77 dipole selection rules 88f spin-SEM 723–6, see also SEMPA spinel (MgA2 O4 ) crystal 220f, 238 diffraction patterns 525f spin–orbit interaction, coupling 76, 77f SPIONs, superparamagnetic iron oxide nanoparticles 594–5 SPLEEM microscopy 26fn sputtering 307, 312 and Auger depth profiling 128f electron microscopy 299–301 ion-beam 313–4 rate 301 SIMS 328–31 threshold energy 300f yield data for aluminum 313f stacking faults 604 dislocations and 656–7f, 658t, 664 Staphylococcus aureus, cell wall digestion by antibiotic 791f static contact mode, topographic imaging 769–71 steel alloys Bessemer process vs open-hearth process 14–15

compositional analysis with dual-phase, AES and XPS 134–5 microstructure 20–1f spectroscopy with Fe and Cr 20–1 electrons 119 see also superalloys spectrum of 2 keV electrons stereographic (probe) 125f projection 244–6 structure determination Stokes/anti-Stokes 485 scattering 166–9, 210 surface electron transitions 180f diffraction 242 stopping power S 315, 316f applications 489–501 strain, macroscale/microscale 461 Low-energy electron diffraction 490–4 Stranski–Krastanov (S–K) reciprocal lattice rods 506 growth 489, 500 Reflection high-energy and island growth 498 electron streaks diffraction 494–501 or relrods in electron surface lattice structures 486f, diffraction 436, 487f 508–9 surface nets 486–7 surface disorder 496 surface reconstruction strength-to-density ratio, 486–7 MSE 3f surface structure, equilibrium structure characterization and growth modes, surface analysis 2 and interface structure factors 505, 507 energies 498f calculations, different crystal surface-enhanced Raman structures 425–30 spectroscopy Fhkl 423 Fridel law 431 (SERS) 183–4 symmetry and 431–2 surface-sensitive signals 89 Strutt, JW 376fn surface-tilt, shadow contrast, Substrate, unreconstructed secondary electron and substrate 489 back-scattered sunflower pollen grains 278fn, images 713f 388 symmetry superalloys 3f elements,inversion, bright field TEM image 11f roto-inversion, HAADF-STEM image 11f screw 240 microstructure analysis 11f and International Tables for and specialty ceramics, Crystallography 239– MSE 3f 244 ultrahigh temperature, for jet structure factor 431 engines 8–10 synchrotron X-ray radiation superconductors, MSE 3f (XRS) 105–6, 138–9, superlattices 464–5 198, 408–81 superparamagnetic iron oxide advanced light source nanoparticles (ALS) 105, 107f (SPIONs) 594–5 groundwater colloids 138–9 superposition principle 348 in situ 468–9 superposition of waves 425 third generation ALS and two waves of same APS 106 frequency 351–3 XAS studies 198 surface analysis 119–37 zone plates 374

T table of values, symbols, SI units and CGS units 825t Tagged Image File Format (TIFF) 50, 51f tapping model (intermittent-contact AFM) 759, 774, 775f, 789 Ta–Si film 128 temporal resolution,defined 32 tenfold rotation symmetry 39f, 266, 481 tensile strength, ultimate (UTS)measurement 20 tetrahedral sites 235 tetrahedron, processing– structure–property– performance 4f thermal field emission guns see Schottky emission guns thermionic emissions 558–9 electron sources 282 from solids, work function 186 sources/cathodes 284t, 558–9 thermionic gun 559, 694–5 thickness fringes CBED patterns 524 diffraction 511, 512, 524, 526f thin foils 511 thin films and multilayers asymmetric arrangement 445 diffractometry, reflectivity and pole figures 445–9f Mo—Si 659f X-ray analysis procedure 645 thin foils 484 bend contours 512 electron transmission 505–6f thickness fringes 511 transmission electron diffraction (TED) 484 two-beam diffraction condition 509–14

OUP CORRECTED – FINAL

844 Index thin lens approximation 378, 562 advances, energy Thompson, Sir JJ 410fn resolution 200 Thompson coherent scattering electron energy-loss 410–13, 504, 536, 541 spectroscopy 623–41 three-window elemental energy dispersive X-ray mapping method 638, spectrometry 641–8 640f energy-filtered threshold energy 299 imaging 639–41 sputtering 300f examples 527–35 Ti indexing a single crystal TiO2Z-contrast image 600f pattern 527 Ti–Al superalloy 128 polycrystalline tilt range 653f materials 527 time of flight (TOF) mass orientation relationships 529 spectrometer 330–2f chemical order 529–31 time scales, dynamic long-period studies 57 multilayers 531–2 time-resolved studies 57 twinning 532–5 neutron-scattering, forward direction 484 hydration of high-energy electron cement 538–40 diffraction 501–3 tip–specimen interactions 767 indexing patterns 548 RMSTitanic, metallurgy failure methods 514–26 analysis 13–15f selected area Toluca iron meteorite 41–2f, diffraction 514–9 42f Kikuchi lines, maps and top-down nanofabrication 738 patterns 519–23 topographic imaging, static convergent beam electron contact mode 769–2 diffraction 523–6 topographical contrast 716 patterns 482, 548 back-scattered electron quasicrystals 267, 481 density summary 816t distribution 715–17 thin foils 484 secondary electron transmission electron imaging 713–15 microscopy torsional spring constant 761 (TEM) 553–692 total reflection XRF applications (TRXF) 113, 114f analysis of defects 655–8 Townes, CH 282fn dynamic trajectory envelopes, point measurements 664–6 incident beam 708f electron tomography transistors (MOSFETs) 660, 650–5 785 semicoductor transition manufacturing: probability,EELS 629 metrology, process transmission electron development & failure aberration-corrected analysis 650–5 microscope thin films and (TEAM) 45 multilayers 658–60 transmission electron basics of electron diffraction 505–9 diffraction (TED) 20, beam-solid interactions and 220, 232, 264, 482, signals 557f 505–9

bend contours 511–2 contrast mechanisms 588–617 Diffraction contrast 595–8 Elastic interactions 589–92 Holography 613–7 Magnetic contrast 609–13 Mass-thickness 592–5 Phase contrast (HREM) 601–09 Z-contrast 598–601 contrast transfer function 579 correction of lens aberrations 587 criteria for technique selection 29–31 cross-sectional view, instrument 554f defects, dislocations and stacking faults 604, 655–6f, 659t, 664 diffraction of electrons in transmission 505–35 digital cameras 514fn electromagnetic lenses 560–5 microscope elements and operations 558–88 electron holography 613–7 first instrument 44, 552, 553, 582fn illumination, axial/off-axis 565–8, 602 images collection of information 553 diffraction modes 573–5f high resolution atomic structure 45f phase contrast imaging, in practice 601–09 recording and detection of electrons 588 through focus series 583f weak beam 658 imaging section 568–70

objective lens and aperture 568–70 resolution 569 instrument and modes of operation 558–88 bright field, dark field and phase contrast imaging 573–84 electromagnetic lenses 560–5 electron sources 282–4t, 558–9 Illuminations section 565–8 Imaging section 568–70 magnification section 571 scanning transmission mode 584–7 specimen handling 570–2 intermediate lens astigmatism 514fn inelastic scattering and spectroscopy 619–23 Lorentz microscopy 609–13 magnification section 571–3 microanalysis EELS 632–7 EDXS 641–8 microdiffraction 648–50 Oxford JEOL JEM2200FS 587f parallel beam operation 565–6 scanning transmission mode and principle of reciprocity 585–7 scattering processes 297 semiconductor process development, failure analysis 660–4, 664f specimen handling and manipulation 570–1 specimen preparation 666–71 summary 671–4, 825t thickness fringes 511–2 see also electron microscopy transmission factor for X-rays 453 transmission high-energy electron diffraction 501–3 tripod scanner 778f

OUP CORRECTED – FINAL

Index 845 tube scanner 777, 778f tungsten carbide pneumoconiosis 137 tungsten tip, FIM image 46f tungsten—halogen lamps 278 tunneling current 756 one-dimensional barrier 749–50 STM 746–8f, 752–3 tunneling spectroscopy 755–7f twin boundary 249, 532 twinning 532–5 diffraction contrast 598f reciprocal lattice 532–3f tip artifacts 779f two-beam diffraction condition 509–14 for DF imaging 596f edge dislocations, bright field imaging 658f thin foil 509–14 two-dimensional structure image 605

U ultimate tensile strength (UTS) 20 measurement 20–1f ultrafast TEM 666 ultrahigh vacuum (UHV) 120 ultramicrotomy 668f, 669 ultraviolet, extreme (EUV) radiations 22f, 23 ultraviolet photoelectron spectroscopy (UPS) 30, 129, 152, 188–95 summary 811t see also X-ray photoelectron spectroscopy (XPS); photoemission spectroscopy (PES) ultraviolet and visible (UV-Vis) spectroscopy 23f, 162–6, 185–6, 812t undulators 105 unmixed indices 426 unoccupied density of states 186f, 199f, 751 UV see ultraviolet

V V, structure 236–7 vacuum chamber 120 vacuum level, defined 185 valence band (VB) 151, 152, 190, 700 energy levels, probed by UPS 194 plasmon resonance 184, 618f valence electrons defined 73, 76f s and p electrons, hybridization 76f shake-up 132 values, table of symbols, SI units and CGS units 825t van de Graaf accelerator 291–2 van der Waals forces/interactions 759f, 765–7, 782 vibrational ellipse, electric vector E 29f vibrational energy levels, diatomic molecule 158f, 158 vibrational/rotational modes 157, 158f, 161f, 162f degrees of freedom 161 virtual excited state 168 von Ardenne, Manfred 584fn von Laue, Max 256fn

W Waller, I 452fn water, vibration modes 162f Watson, James 12fn wave equation for simple harmonic motion 346 wave—particle duality 24, 304 de Broglie 26, 69 wave—vibrational/rotational modes, particle duality 304 wavelength electromagnetic spectrum 23f, 37, 172

and energy dispersive X-ray spectra 109f incident radiation 38 of light, defined 346 short wavelength limit (SWL) 99, 101 X-ray intensity as function of 99f wavelength dispersion X-ray spectrometry (WDXS) 36, 107–10f, 145, 437–8 compared with energy-dispersive X-ray spectrometry (EDXS) 109f, 112 diffracting crystals 109t high resolution and high signal to background 109 summary 812t wavenumbers, bending and stretching vibrations 172t waves, subtraction of amplitudes 38f weak phase object (WPO) approximation 577 Wehnalt electrode 285, 286f white lines 3d transition metals 200 Cu 200 intensity L3/L2 203f transitions 201f Wien filter 289f Wilkins, Maurice 12f work function 97 conversion into SI units 25 defined 186 Wulff net 245–6f Wurtzite 237f Wycoff letter 243

X X-ray absorption and filtering 102–3 X-ray absorption near edge structure (XANES) 187, 199–200, 208–10f, 630, 812t X-ray absorption spectroscopy (XAS) 197–206

analysis of catalytic particles 208–9 energy ranges of interest 199f L-edge spectra of tetrahedrally coordinated Si 198f near-edge (NEXAFS)and extended fine structure (EXAFS) 199–206, 210 single-electron transfer picture 198 summary 812t total electron yield detection 200 unfilled levels 151 and XPS 187f X-ray corelevel spectroscopy 107–118 X-ray diffraction pattern (è –2è scan)function of angle 39f X-ray diffraction (XRD) 248–77, 257, 361–3, 408–80 applications 459–71, 468–9 Bragg–Bretano geometry 440f chemical order-disorder studies 464–7 by a circular aperture or disc 372–3 comparison of X-ray and electron diffraction 262–5 Ewald and limiting sphere construction 260f factors influencing intensities 451–3 Fraunhofer and Fresnel diffraction 362f, 370–3 Huygens’s principle 356 in situ 468–9 measurements on Mars 469–80 powder diffraction files (PDF) 461–2f short-range order and diffuse scattering 467–8 summary 471–2, 817t

OUP CORRECTED – FINAL

846 Index X-ray diffraction (cont.) systematic absences 431 temperature dependence 466 see also area —; convergent beam electron —; low-energy electron —; transmission electron— X-ray emission lines 89t, 100f all elements 89t X-ray emissions, nomenclature 85f X-ray fluorescence spectroscopy (XRF) 78, 89, 97, 113, 137–9, 410 analysis of dental and medical specimens 137–8 application in environmental science 138–9 microfocusing, medical applications 138f paintings 12–13 photon incidence 25, 113 summary 813t total reflection XRF 113, 114f X-ray generation 98–107 X-ray intensity, function of wavelength, Mo X-ray tube 99f X-ray magnetic circular dichroism (XMCD) 43 X-ray microanalysis 726–7 absorption or transmission factor 453 Al-Cu alloy 730

atomic number (Z) correction 728 atomic scattering factor 414–16 attenuation, computed tomography 16 chemical microanalysis 113 Compton incoherent scattering 413–14 electromagnetic spectrum 23f electron probe microanalysis (EPMA) 16, 78, 89, 114–19, 618 fluorescence, following creation of core hole 36f generation 97–107 hard/soft X-rays 23 high angle scan, Cu K-alpha radiation 446 neutrons, electrons, scattering factors 536t production with depth2 specimens of different densities 115f quantitative X-ray microanalysis 116 in a SEM calibration curves for Fe-Ni binary alloy 727f error analysis 729 fluorescence (F) correction 728 Thomson coherent scattering 58, 410–13, 421, 504, 536 wavelength dispersion spectrometer (WDS) 107–8f

X-ray absorption (A) correction 728 X-ray microscopy 43 zone plates 373–4 X-ray photoelectron spectroscopy (XPS) 22, 31f, 69, 78, 128–36, 140, 188f –97 Al K-alpha radiation 218 comparison of AES and XPS 136–7 electrostatic deflection analysers 123 examples of chemical shifts 132f filled levels 151 partially oxidized surface of Al with contamination 130f summary 813t surface sensitivity 190f vacuum chamber 120 X-ray photons core-shell vacancies 107 detection 107–12 dipole selection rules 91 X-ray shielding 145 X-ray take-off angle 619f X-ray transmission microscopy (XTM) 34, 43f summary 823t X-ray tube 97–8f Ag as target material 100 efficiency 101 principal components 100f short wavelength limit (SWL) 99

X-rays as probes see X-ray fluorescence spectroscopy (XRF) X-rays as signalsEPMA 16, 78, 89, 114–19 X-ray absorption near edge structure XANES 630 xenon emission profile 279f manipulation of adsorbed atoms on clean surfaces 757 XPS/ESCA spectrum 145

Y YBa2 Cu3 O4 spectrum 323 YBa2 Cu3 O7 back-scattering electron yield 713 Young, Thomas 358fn Young double-slit experiment 357–9f Young modulus 761fn, 762

Z Z-contrast imaging 555, 557f, 598–601 summary 823t ZAF correction factors 117–18 zero order Laue zone (ZOLZ) 263, 525f zero point energy 159 zinc blende, structure 238, 244 zone half-period 371–3 zone axis, common line 232f zone plates, X-ray microscopy 43, 373–4f

Name

Symbol

Value

SI units

Avagadro’s Number

NA

6.022 × 1023 mol−1

—

CGS units —

Boltzmann’s Constant

kB

1.38062

×10−23

Bohr Magneton

μB

9.274

×10−24 Am2 or JT−1

×10−21 erg G−1

Bohr’s radius for H

a0

5.292

×10−11 m

×10−9 cm

Electron Charge

e

1.6022

×10−19 C

4.8032 ×10−10 esu

Electron rest mass

me

9.109

×10−31 kg

×10−28 g

Electric permittivity of free space

ε0

8.854

×10−12 F m−1

1

Magnetic flux quantum

0 = h/(2e)

2.0678

×10−15 Tm2

Magnetic permeability of free space

μ0

4π × 10−7 [Hm−1 ]

1

Planck’s constant

h

6.626

×10−34 J

s

×10−27 erg s

Reduced Planck’s constant

h¯ = h/2π

1.0546

×10−34 J s

×10−27 erg s

Proton rest mass

mp

1.6726

×10−27 kg

×10−24 g

Velocity of light in free space

c

2.9979

×108 m s−1

×1010 cm s−1

1 electron volt

eV

1.6022

×10−19 J

×10−12 erg

Atomic mass unit

amu

1.6606

×10−27 kg

×10–24 g

J

K−1

×10−16 erg K−1

OUP CORRECTED – FINAL

Table of Values

IA

Periodic Table of the Elements

VIIIA

1

2

1.01 2.1 13.595

4.00 –– 24.58

Hhcp

Hehcp

IIA

1s 3

Key

4

Li bcc Be hcp 2s 2

2s 11

12

22.99 0.9 5.14

24.31 1.2 7.64

24

Symbolcrystal structure

52.00 1.6 6.76

Atomic Weight

Cr bcc

6.94 9.01 1.0 1.5 5.39 9.32

IIIA IVA Atomic Number

3d54s

Na bcc Mg hcp

Electronegativity (Pauling) Energy(eV) to remove one electron Electron (outer) configuration

Intermediate Metal

13

39.10 0.8 4.34

40.08 44.96 47.87 50.94 52.00 54.94 55.85 58.93 58.69 1.0 1.3 1.5 1.6 1.6 1.5 1.8 1.8 1.8 6.11 6.56 6.83 6.74 6.76 7.43 7.90 7.86 7.63

22

Rb bcc Srfcc

85.47 87.62 0.8 1.0 4.18 5.69

5s 55

5s 2 56

Csbcc Babcc

132.91 137.33 0.7 0.9 3.89 5.21

6s 2 88

(223) 0.7 ––

(226) 0.9 5.28 7s 2

Fr 7s

Rabcc

24

26

25

27

28

15

14

9

1s 2 10

19.00 4.0 17.42

20.18 –– 21.56

Orhomb Fmono

16

Nefcc

2s 22p5 2s 22p6 17 18

29

IIB 30

26.98 28.09 30.97 1.5 1.8 2.1 5.98 8.15 10.55

32.06 35.45 39.95 2.5 3.0 –– 10.36 13.01 15.76

3s 23p 3s 23p2 3s 23p3 3s 23p4 3s 23p5 3s 23p6 31 33 34 35 36 32

63.55 65.41 69.72 72.64 74.92 78.96 79.90 83.80 1.9 1.6 1.6 1.8 2.0 2.4 2.8 –– 7.72 9.39 6.00 7.88 9.81 9.75 11.84 14.00 3d104s 3d104s 2 4s 24p 4s 24p 2 4s 24p 3 4s 24p4 4s 24p 5 4s 24p 6

3d54s 2 3d64s 2 3d74s 2 3d84s 2 43 44 45 46 47

48

49

51

50

52

Yhcp Zrhcp Nb bcc Mo bcc Tchcp Ruhcp Rhfcc Pd fcc Agfcc Cdhcp In fct Sndiam SbRhom Tehex

53

I ortho

54

Xefcc

88.91 91.22 92.91 95.94 (98) 101.07 102.91 106.4 107.87 112.41 114.82 118.71 121.76 127.60 126.90 131.30 1.2 1.4 1.6 1.8 1.9 2.2 2.2 2.2 1.9 1.7 1.7 1.8 1.9 2.1 2.5 –– 6.5 6.95 6.77 7.18 7.28 7.36 7.46 8.33 7.57 8.99 5.78 7.34 8.64 9.01 10.45 12.13

4d5s 2 4d 25s 2 4d 35s 2 4d45s 2 4d55s 2 4d65s 2 4d75s 2 4d85s 2 4d105s 4d105s 2 5s 25p 5s 25p2 5s 25p3 5s 25p4 579 80 82 72 73 74 75 76 77 78 81 83 84

Hf hcp Ta bcc Wbcc Re hcp Os hcp

Irfcc Pt fcc

Aufcc Hgrhom Tlhcp Pbfcc Birhom Pocub Atfcc Rnfcc

178.49 180.95 183.84 186.20 190.23 192.20 195.08 196.97 200.59 204.38 207.19 208.98 (209) 1.3 1.5 1.7 1.9 2.2 2.2 2.2 2.4 1.9 1.8 1.9 2.0 1.8 7.0 7.88 7.98 7.87 8.7 9.0 8.96 9.22 10.43 7.41 7.29 8.43 6.11 4f 145d 26s 2

57

5d36s 2 5d46s 2 5d56s 2 5d 66s 2 58

59

60

61

5d 9

5d 96s

62

63

5s 25p5 5s 25p6 85 86

5d106s 5d106s 2 64

65

6s 26p

6s 26p2

66

67

6s 26p3 6s 26p4 68

(210) (222) 2.2 –– –– 10.74

6s 26p5 6s 26p6

69

70

71

Lahcp Ce fcc Prhcp Nd hcp Pmhex Smrhom Eubcc Gd hcp Tb hcp Dyhcp Hohcp Erhcp Tmhcp Ybfcc Luhcp

138.91 140.12 140.91 144.24 (145) 150.35 151.96 157.25 158.92 162.5 164.93 167.26 168.93 1.1 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 5.61 6.91 5.76 6.31 –– 5.60 5.67 6.16 6.74 6.82 6.01 6.10 6.18

5d6s 2 4f 26s 2 89 90

173.04 174.97 1.2 1.2 5.43 6.25

4f 36s 2 4f 46s 2 4f 56s 2 4f 66s 2 4f 76s 2 4f 86s 2 4f 96s 2 4f 106s 2 4f 116s 2 4f 126s 2 4f 136s 2 4f 146s 2 4f 145d6s 2 –– 91 92 93 94 95 96 97 98 99 100 101 102 103

Ac fcc Thfcc Pa bct Uortho Nportho Pumono Amhcp Cmhcp Bkhcp Cfhcp Eshcp

Fm

(227) 232.04 231.04 238.03 (237) (244) (243) (247) (247) (251) (252) (257) 1.1 1.3 1.5 1.7 1.3 1.3 1.3 1.3 1.3 1.3 1.3 1.3 5.17 6.08 5.89 6.05 6.19 6.06 5.99 6.02 6.23 6.30 6.42 6.50 2 2 6 2 7 2 7 2 2 2 3 2 4 2 2 –– –– –– –– 6d7s 6d 7s 5f 6d7s 5f 6d7s 5f 6d7s 5f 7s 5f 7s 5f 6d7s

Md

No

(258) (259) 1.3 1.3 6.58 6.65 –– ––

Lr

(262) 1.3 –– ––

OUP CORRECTED – FINAL

6s 87

B rhom Cdiam Nhcp

8

Vbcc Cr bcc Mn bcc Fe bcc Cohcp Ni fcc Cufcc Zn hcp Gaortho Gediam Asrhom SeMon Brortho Krfcc

3d4s 2 3d24s 2 3d34s 2 3d54s 39 40 41 42

RARE EARTHS

4s 2 38

ACTINIDES

4s 37

23

7

Al fcc Sidiam Portho Sortho Clortho Arfcc

IB

VB VIB VIIB

3s 2 20

21

VIA VIIA

10.81 12.01 14.01 16.00 2.5 2.0 3.0 3.5 8.30 13.61 11.26 14.54 2s 22p 2s 22p2 2s 22p3 2s 22p4

3s 19

K bcc Ca fcc Sc hcp Ti hcp

VA

Nonmetal

VIII IIIB IVB

6

5